Professional Documents
Culture Documents
CHAPTER – 2
REVIEW OF LITERATURE
2.1 GENERAL
The mechanical properties and service-life of components evoked
global market for lightweight components having greater strength and stiffness.
Composite materials have the potential to replace widely used metals or their
alloys such as steel, aluminum and their alloys. It is revealed that 60% to 80%
components, while 20% to 50% weight saving occurs if aluminum parts are
many available composite material systems are quite complex, costly and time
composite materials along with required properties are summarized in Table 2.1.
14
al, 1992; Yue et al, 1998; Engine cylinder liners (Al-SiC), CNG Improved wear
Peters, 1998; Fitzpatrick et storage cylinders, Diesel Engine pistons resistance, Thermal
al, 1998; Hunt, 2000; (SiCw/Al-alloy), Brake rotors, Leaf fatigue resistance,
Surappa, 2003; Bayat et al, springs (E-glass/epoxy), Drive shafts Weight reduction, High
2007; Singh, 2008; Bayat (Al-C), Flywheels, Racing car brakes thermal conductivity
Sub-Marine (Peters, 1998) Propulsion shaft (Carbon and glass Weight reduction,
Commercial and Industrial Computer hard disk drive, Needle for Weight reduction,
Beams.
Aerospace equipment and Rocket nozzle (TiAl-SiC fibers), Heat Light weight, Specific
structures, Space structure exchanger panels, Engine parts (Be-Al), stiffness and specific
(Yue et al, 1998; Pitcher et Wind tunnel blades, Spacecraft truss strength at elevated
al, 1998; Bache et al, 1998; structure, Reflectors, Solar panels, temperature, Creep and
Peters, 1998; Hunt, 2000) Camera housing, Hublle space telescope fatigue resistance,
Aircraft, Missile structures Wings, Rotary launchers, Engine casing, Stiffness, Reduced
Shakesheff and Purdue, Propeller blades, Landing gear doors, temperature stability
(Carbon/Epoxy).
Nuclear Reactor (Hunt, Storage casks for spent fuel rods from Absorption of neutron
Sports (Fitzpatrick et al, Tennis rackets, Golf shafts, Racing Increased specific
1998; Hunt, 2000) bicycle frame (SiCw/6061), Fishing rod, stiffness and strength,
The yield criteria of materials limit the elastic domain during loading,
whereas the failure criteria give the maximum stress that can be applied.
Traditionally, we use the term yield criteria for metals or alloys and failure criteria
for geomaterials such as soil and concrete. Safe and efficient use of materials is
required for the successful design of any structural components. Therefore, for
design purposes, the onset of plastic yielding under loading conditions is of great
importance. The yield criterion gives the onset of plastic deformation. In other
words, if a state of stress satisfies yield criterion, one can say that the plastification
may start. It is assumed that the initial yielding is dependent only on the state of
stress and not on how the stress is reached. It can be assumed that there exists a
As the yield criterion does not depend on the path of loading, it does not tell
multiaxial stress state are combined into a single quantity known as the effective
stress (σ e ) . The effective stress is then compared with the yield stress in some
The state of stress can be determined by specifying the principal stress and
orientation of principal axes. The three principal stresses and three angles
constitute the six dimensional spaces. Thus, the yield function can be written as,
f = f (σ 1 , σ 2 , σ 3 , α 1 , α 2 , α 3 )
principal axes.
observations, the chief of which is that pure hydrostatic pressure should not cause
There exist three principal stresses and their orientation for a given state of
stress at a point. Hence one can represent the state of stress as a point in three
dimensional vector space whose bases are the principal stresses for isotropic
materials, where the orientation of the axes is not important. For isotropic materials
acted upon by deviatoric stress, then the yield surface can be represented as,
f (σ 1 , σ 2 , σ 3 ) = 0 (2.1)
σ1 + σ 2 +σ 3 = 0 (2.2)
the principal axes. This plane is called the π - plane. Equations (2.1) and (2.2)
independent of hydrostatic stress, then the yield function is a right cylinder with
orientation of principal axes i.e. α 1 , α 2 and α 3 . Thus, in this case the yield function
f = f (σ 1 , σ 2 , σ 3 )
It was proposed by von Mises (1913) that the yielding would occur in an
isotropic material when the second invariant of stress deviator J2 exceeds some
critical value.
J2 =k 2 (2.3)
where,
J2 =
1
6
[
(σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 ]
To estimate the value of constant k and to relate it with the yielding under tension
σ e2 + σ e2 = 6k 2
or, σe = 3k (2.4)
Using Eqn. (2.4) into Eqn. (2.3), we obtain the following usual form of von Mises
yield criterion,
σe =
1
[(σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 ]
1/ 2
(2.5)
2
whose generator is equally inclined to the principal stress axes, as shown in Fig.
2.2. The von-Mises criterion implies that the yielding under both uniaxial tension
and compression would start at the same value of tensile and compressive stresses.
Arsenault and Taya (1987) pointed out that even in an isotropic metal
matrix composite yielding does not begins at the same level of tensile and
compressive stresses under uniaxial loading. Badini (1990) also noticed that the
yield strength in tension. The processing of metal matrix composites often involves
cooling from the higher temperature, which results in residual tensile thermal stress
(Schellekens and De Borst, 1990a, b; Bicanic et al, 1994; Moin, 1996) employed
(σ 12 + σ 22 + σ 32 ) − (σ 1σ 2 + σ 2σ 3 + σ 3σ 1 ) + Q(σ 1 + σ 2 + σ 3 ) − 1 = 0 (2.6)
20
the following alternate form of Hoffman’s yield criterion that employs uniaxial
(σ 12 + σ 22 + σ 32 ) − (σ 1σ 2 + σ 2σ 3 + σ 3σ 1 ) + ( f c − f t )(σ 1 + σ 2 + σ 3 ) − f c f t = 0
(2.7)
orientation of whiskers in the matrix (Ledrich and Shastry, 1982; Crowe et al,
1985; McDenels, 1985). Badini (1990) observed that the compressive yield
criterion given by Eqn. (2.5) is not an appropriate choice for such cases.
which show different yield stresses under tension and compression. Several
investigators (Arsenault and Taya, 1987; Shi et al, 1992) reported that thermal
and tension is dependent on the test temperature, as reported in Table 2.2 for 20
The yield stresses reported in the table above are deduced from the stress-
Badini (1990) studied the correlation between the microstructure and the
tensile and compressive properties of extruded bars made of 6061 Al alloy matrix
the longitudinal direction was observed to be considerably higher than the strength
samples, having 20 mm diameter, was found to be 218 MPa and 186 MPa
direction of load applied (Lederich and Sastry, 1982; Crowe et al, 1985; McDanels,
1985).
22
2.3 CREEP
which is a function of time and increases continuously even under constant load.
generally above 0.5 Tm, results in permanent deformation even when the stress is
well below the yield stress. This deformation in metals or metal matrix composites
depends on the effective stress through a power law involving material parameters,
(1980) pointed out that the creep behavior is a function of stress, temperature, time,
stress history, temperature history and type of material. Creep behavior also
structure with time while the total strain remains constant. Further, it also includes
the recovery, which is characterized by the reduction of inelastic strain with time
nuclear power plants, missiles, aeroengines, gas turbines etc. The testing and
The increase in stress and temperature increases the creep rate and reduces
the rupture time. Creep is generally viewed as a thermally driven process that is
glide, bulk diffusion, grain boundary diffusion and dislocation creep have been
conceived. More than one mechanism could operate at a given temperature. Ashby
and Frost (1975) developed a convenient summary of the various active creep
maps shown in Fig. 2.3. The deformation maps depend also on the microstructure
of material. Increase in grain size reduces the amount of grain boundary diffusion
specimen is heated to a constant temperature (T) of the order of 0.3 to 0.5Tm (Tm is
the melting temperature of the material) and subjected to a constant tensile force F.
The strain developed in the specimen is recorded and represented with respect to
24
time, in order to get creep curve of the material. The normal stress induced in the
specimen is usually much less than the yield limit of the material.
Curve A in Fig. 2.4 illustrates the idealized shape of a creep curve. The
variation of creep strain with time. The slope of curve A, (dε / dt or ε& ) is referred to
as creep rate. The degree to which the three stages of creep are readily
distinguishable depends strongly on the applied stress and temperature. The first
rate. Primary creep is a period of predominantly transient creep during which the
creep resistance of material increases by virtue of its own deformation. For low
creep is the predominant creep process. The second stage of creep, known also as
secondary creep, is a period of nearly constant creep rate, which results from a
balance between the competing processes of strain hardening and recovery. For this
reason, the secondary creep is usually referred to as steady state creep. The average
value of the creep rate during secondary creep is called the minimum creep rate.
Third-stage or tertiary creep mainly occurs in constant load creep tests at high
stresses and high temperatures. Tertiary creep occurs when there is an effective
formation. Third stage creep is often associated with metallurgical changes such as
phases that are present. The dashed line (curve B) in Fig. 2.4 shows the shape of a
found that the onset of stage III is greatly delayed. In engineering situations it is
usually the load not the stress that is maintained constant, so a constant-load creep
Andrade’s (1957) showed that the constant stress-creep curve represents the
superposition of two separate creep processes, which occur after the sudden
straining due to application of load. The first component of the creep curve is a
transient creep in which the creep rate decreases with time. Added to this is a
is shown in Fig. 2.5. Andrade observed that a creep curve can be represented by the
ε = ε 0 (1 + β 1t 1 / 3 ) e kt (2.8)
where ε 0 is the instantaneous strain, ε is the strain in time t, and β 1 and k are
constants.
models of creep are needed. For this purpose, the data obtained from the tensile
If the slope of a creep curve (Fig. 2.4) is plotted versus strain, a curve of
creep rate vs total strain is obtained (Fig. 2.6). This curve illustrates the large
change in creep rate, which occurs during the creep test. Since the stress and
temperature are constant, this variation in creep rate is the result of changes in the
various locations in a creep specimen have shown that the local strain undergoes
many periodic changes with time that are not reflected while recording the changes
in strain over total gage length of the specimen. In large grained specimens, local
Most of the models representing uniaxial creep of metals and alloys are
widely used and implemented as options in several finite element codes. The
models express uniaxial creep strain or creep strain rate as a function of stress, time
and temperature, and some models also use the accumulated creep strain to model
ε = f (σ , t , T ) (2.9)
where ε is the creep strain, σ is the applied stress, t is the time period and T is the
temperature of application.
ε = f 1 (σ ). f 2 (t ). f 3 (T ) (2.10)
Since the temperature during creep test is constant, therefore, Eqn. 2.10
becomes,
ε = f1 (σ ). f 2 (t ) (2.11)
27
developed in the past. The most commonly used stress function law is Norton’s law
given as,
f1 (σ ) = Bσ n
where B and n are the material constants. The approximation of a creep curve to a
well as primary creep is negligible. Under such conditions, it can be assumed that
the creep strain rate depends upon stress function only. The time dependence of
creep has been expressed in terms of Bailey’s empirical law given by,
f 2 (t ) = A1t m
temperature and stress over the entire life of component. Therefore, it is not
possible to compute the creep strain directly from a uniaxial creep equation. The
rules of time hardening and strain hardening have been developed to extend the
temperature histories. The creep strain is determined by integrating the creep strain
rate equations for changing stress and temperature conditions. To compute the
creep strain rate at a particular instant, the mechanical and thermal loads as well as
the material history must be known. Hardening rules specify the state of material as
looked upon as a method of moving between constant stress and temperature creep
curves as the stress and temperature change. Two different hardening rules have
28
been used with classical creep models: time hardening and strain hardening. When
the creep rate equations are integrated for conditions of constant stress and
temperature, both the hardening rules produce identical results, namely the uniaxial
The time hardening theory states that for a constant temperature and
variable stress condition, creep rate ( ε&c ) is a function of stress and time i.e.,
ε&c = f (σ , t ) (2.12)
However, in case of strain hardening theory it is assumed that the creep rate
ε&c = f (σ , ε ) (2.11)
The particular forms of these laws can be obtained by assuming that the
representation of creep in the primary and secondary creep ranges under isothermal
ε c = Aσ n t m (2.12)
where A, m and n are constants whose values depend upon the type of material. The
The form of Bailey Norton given by Eqn. (2.12) could be used to describe creep in
both tension and compression, but it would be reasonable to expect that the
Differentiating Eqn. (2.12) with respect to time, the time hardening law can
be obtained as,
29
dε c
ε&c = = Amσ n t m −1 (2.13)
dt
From the above equation it can be seen that the creep rate decreases with
time since 0 < m < 1. Further, Eqn. (2.13) can also be written in the following form,
mA (1 / m )σ ( n / m )
ε&c = (2.14)
ε c(1− m ) / m
Eqn. (2.14) indicates that the creep strain rate decreases with increasing creep
strain ( ε c ) i.e. with the progression of creep strain the material hardens.
Though, both the laws are derived from the same equation, but it is
observed that for varying stress, the time and strain hardening laws give different
creep rates. This difference is procedural and not phenomenological. Quite often
the strain hardening models give more accurate predictions of experimental results
always yield accurate predictions, particularly when several step changes in stress
occur during the same test (Rabotnov, 1969). Pickel et al (1971) also noticed that
the strain hardening model is unable to accurately predict the creep behavior
resulting from structural instabilities. But for structurally stable materials, the
predictions by strain hardening model are fairly reliable. However, in the case of
gradually varying stress, both the laws give approximately similar predictions.
The uniaxial creep tests allow us to establish the basic features of creep
behavior and to establish the relationship between stress, strain rate, temperature,
time, etc. However, in actual practice, most of the structural components operate
30
under multiaxial stress creep are limited and the available models of multiaxial
creep are mainly phenomenological in form (Kraus, 1980; Boyle and Spence, 1983;
Gooch and How, 1986). Therefore, in order to analyze the influence of stress state
given below,
ε&e ⎡ 1 ⎤
ε& x = ⎢σ x − 2 (σ y + σ z )⎥
σe ⎣ ⎦
ε&e ⎡ 1 ⎤
ε& y = ⎢⎣σ y − 2 (σ z + σ x )⎥⎦
σe
ε&e ⎡ 1 ⎤
ε& z = ⎢σ − (σ x + σ y )⎥
σe
z
⎣ 2 ⎦
3ε&e
ε& zx = τ zx
σe
3ε&e
ε& xy = τ xy
σe
3ε&e
ε& yz = τ yz (2.15)
σe
ε& xy , ε& yz , ε& zx are respectively the normal and shear strain rate components in x, y and
31
z directions respectively. The effective stress ( σ e ) and the effective strain rate ( ε&e )
σe =
1
[(σ y − σ z ) 2 + (σ z − σ x ) 2 + (σ x − σ y ) 2 + 6(τ yz2 + τ zx2 + τ xy2 ) ] 1/ 2
(2.16)
2
ε&e =
1
[(ε& y − ε& z ) 2 + (ε& z − ε& x ) 2 + (ε& x − ε& y ) 2 + 6(ε& yz2 + ε& zx2 + ε& xy2 ) ]
1/ 2
(2.17)
2
The Eqs. (2.15) - (2.17), are based on the assumption that the material is
isotropic.
The steady state creep dominates at temperatures above about 0.5Tm. The
simplest assumption that the creep is a thermally activated process, which can be
ε& s = Ae − Q / RT
the flow unit, the entropy change and the factor that depends on the structure
of the material.
energy for creep. If the temperature interval is small such that the creep mechanism
and,
R ln (ε&1 / ε& 2 )
Q=
(1 / T2 − 1 / T1 )
An extensive correlation between creep and diffusion data for pure metals
(refer Fig. 2.7) indicates that the activation energy for high temperature creep is
equal to the activation energy for self diffusion. The activation energy for self
diffusion is the sum of energies for the formation and movement of vacancies,
which strongly supports the view that dislocation climb is the rate controlling step
another factor in support of this view. Therefore, it is expected that the metals in
which the vacancies move rapidly would have better creep resistance.
creep rate ( ε&e ) is related to the effective stress ( σ e ) through the following well
documented threshold stress ( σ o ) based creep law (Park et al, 1990; Mishra and
Pandey, 1990; Mohamed et al, 1992; Pandey et al, 1992; Gonzalez and Sherby,
1993; Pandey et al, 1994; Park and Mohamed, 1995; Cadek et al, 1995; Li and
Mohamed, 1997; Li and Langdon, 1997a, 1999a; Yoshioka et al, 1998; Tjong and
n
⎛σ −σo ⎞ ⎛−Q⎞
ε&e = A ⎜ e
'
⎟ exp⎜ ⎟ (2.18)
⎝ E ⎠ ⎝ RT ⎠
33
where the symbols A’, n, Q, E, R and T denote respectively the structure dependent
The threshold stress based creep law given by Eqn. (2.18) may alternatively
be expressed as,
ε&e = [M (σ e − σ o )] n (2.19)
1/ n
1⎛ −Q⎞
where, M = ⎜ A ' exp ⎟
E⎝ RT ⎠
the type of material and operating temperature (T). In a composite, the dispersoid
size (P) and its content (V) are the primary variables affecting these parameters.
Therefore, these parameters are functions of dispersoid size (P), volume content of
the dispersoid (V) and operating temperature (T). But the functional relations,
σ o can be extracted from the experimental creep results reported for aluminum or
It is evident from Fig. 2.8 that the variation of creep rate with applied stress
for the composites generally exhibits curvature at higher creep rates when creep
strain rates are measured over more than five orders of magnitude. The apparent
creep does not occur. This similarity indicates that the creep behavior of these
34
creep (Mohamed, 1998) and the creep behavior of these composites may be
Nardone and Strife (1987) were the first to introduce the threshold stress
into power law creep equation, to explain the creep data obtained for 20 vol%
and Pandey et al (1992) also considered the threshold stress while analyzing creep
data of SiCp/Al composites. Using threshold stress based approach, one could
easily explain the high values of apparent stress exponent and activation energy. In
the creep behavior of these composites was rationalized by using the threshold
stress approach used by numerous workers (Nardone and Strife, 1987; Park et al,
1990, 1994; Pandey et al, 1992; Mohamed, et al, 1992; Gonzalez and Sherby,
1993; Cadek et al, 1994, 1995, 1998a, b; Zhu et al, 1996; Li and Mohamed, 1997;
Li and Langdon, 1997a, 1998a, b; Tjong and Ma, 1999a; Ma et al, 1999;
using linear extrapolation of the strain rate versus stress plot in which the creep
Further, it is assumed that if the creep data of composites satisfy Eqn. (2.18), and
35
σ 0 is independent of the applied stress, the data points in ε&1 / n versus σ plot would
fit in a straight line. This line is extrapolated to zero strain rate to get the value of
threshold stress ( σ 0 ) at the given temperature. The most appropriate value of true
data by using different values of n, and selecting the value of n that describes the
described by modified power law given by Eqn. (2.18), the relation between the
apparent stress exponent na, true stress exponent n, applied stress σ and threshold
nσ
na = (2.20)
(σ − σ 0 )
estimated from the data in the plot of apparent stress exponent na versus applied
n which are related to some other well documented creep process. However, this
and Langdon (1997c). When the creep data covers at least 5 orders of magnitude of
36
strain rates and includes the strain rate as low as ~10-8 s-1, the curves through the
vertical at a strain rate of ~10-10 s-1. Accordingly, the predicted stress level at this
low strain rate is very close to the threshold stress. Thus, using the values of
threshold stress, estimated at a strain rate of ~10-10 s-1, the creep data could be
replotted on the logarithmic scale between the measured strain rate against the
normal stress, threshold stress under normal loading, applied shear stress and
threshold stress in shear respectively. The slope of logarithmic plots of strain rates
estimated from ε& 1 / 8 versus σ plot by using linear extrapolation technique are
primarily dependent upon the volume fraction of the reinforcement but are
suggested that the load transfer to the reinforcement may be responsible for the
origin of threshold stress. However, this approach fails to account for the
Mohamed, 1995).
composites, based on the value of true stress exponent (n = 8), Gonzalez and
Sherby (1993) reported that the threshold stress exhibits two main characteristics.
range between 733 K and 743 K. Secondly, the threshold stress is higher for
vol% SiCp/2124Al composite at temperatures between 623 K and 748 K. The plot
of shear creep rate versus applied shear stress ( ln γ& m vs ln τ ) indicates the origin of
threshold stress with the value of apparent stress exponent, na = ( ln γ& m / ln τ )T,
threshold stress into analysis, it is observed that the minimum creep rate of the
composite is controlled by the matrix lattice diffusion with a true stress exponent
735 K. This finding is in good agreement with that of Gonzalez and Sherby (1993),
though these workers have introduced the structure invariant model with a stress
exponent of 8.
38
into fine SiC particles, with sizes smaller than 100 nm, as a result of thermo-
mechanical processing. These fine SiC particles, like oxide particles, partly account
for the origin of threshold stress. Moreover, it is reasonable to assume that high
volume fraction of reinforcement gives rise to a larger number of fine SiC particles.
Thus, the increase in threshold stress is expected with the increase in volume
and 678 K, and compared the results obtained with those noticed for unreinforced
2124Al, tested under similar experimental conditions. The study indicates that the
origin of anomalous stress dependence of the creep rate of the composite and the
samples suggests that the most probable source of threshold stress in the composite
dispersed particles, which includes those formed during the processing of these
materials through powder metallurgy route. The creep behavior of the composite is
During analysis of creep data, the value of true stress exponent (n)
well-documented creep cases for metals and alloys: (i) n = 3 for creep controlled by
temperature dislocation climb (lattice diffusion), and (iii) n = 8 for lattice diffusion-
SiCp/Al (Pandey et al, 1990, 1992) and TiB2p/Al (Pandey et al, 1994) composites.
The creep data obtained has been explained by using the substructure invariant
during creep in these composites the subgrain boundaries are pinned by reinforced
particles like Al2O3 and TiB2, thereby yielding a stress exponent of 8 and an
remains stable during extended creep exposure and such a structure is insensitive to
stress. This is achieved by the introduction of second phase particles such as oxide
boundaries and the subgrain size becomes equal to interparticle spacing. In such a
situation, particles stabilize the subgrain size. The substructure invariant model
predicts that the creep rate is proportional to the cube of subgrain size and a stress
with stress. The substructure invariant model (Sherby et al, 1977) is described by
3
⎛ D ⎞ ⎛ λ ⎞ ⎛ σ − σ ⎞8
ε = S ⎜ L2 ⎟ ⎜⎜ ⎟⎟ ⎜
& 0
⎟ (2.21)
⎜ b ⎟ br ⎝ E ⎠
⎝ r ⎠⎝ ⎠
where ε& is the strain rate, σ is the applied stress, DL is the lattice self-diffusion
Gonzalez and Sherby (1993) reanalyzed the creep data of a series of SiC/Al
composites and demonstrated that the creep in SiC/Al composites may be described
by using the substructure invariant model. They also pointed out that the barrier
Mishra and Pandey (1990) analyzed the steady state creep data of SiC/
al (1988), and noticed that the substructure invariant model (n = 8) could explain
the entire set of data. It is observed that the basic steady state creep mechanism
remains unaltered for different shapes and volume fractions of SiC reinforcement,
that the stress exponent of 8 gives the best linear fit between ε& 1 / n and σ for this in-
situ composite. However, for n = 3 and 5, the data points of the plots exhibit a clear
curvature at low applied stress. It is further noted that the data points for all the
temperatures may be well fitted by a single straight line with a slope of 8, which
41
implies that the substructure invariant model provides satisfactory explanation for
may also be reasonably explained by means of the substructure invariant model (Ma
composite demonstrated that a stress exponent of 5 rather than 8 exhibits the best
linear fit between γ&1s / n and τ , where γ& s and τ denote the shear creep rate and the
composite also revealed that the value of true exponent (n) ≈ 5 and a true
activation energy for creep (Q) ≈ 208 kJ/mol. Therefore, the prevailing creep
(Park et al, 1990), 20 vol% SiCp/Al (Pandey et al, 1992), 26 vol% Al2O3/Al–
5%Mg (Dragone and Nix, 1992) and 30 vol% SiCp/Al (Cadek et al, 1994b)
composites and concluded that the relationship between ε& 1 / n and σ in these
the linear fitting of creep data obtained for SiCp/Al by Pandey et al (1992) could be
achieved satisfactorily for various values of stress exponent (n = 3, 5 and 8), due to
relatively narrow range of experimental creep rates, which are less than four orders
of magnitude.
42
Mg composite and noticed that the composite exhibits a stress exponent (n) = 8 and
higher than the activation energy for lattice self-diffusion. Li and Langdon (1997b)
during re-appraisal of the same data noticed that n ≈ 3 and Q ≈ 125 kJ / mol , which
is consistent with the creep controlled by a viscous glide process and the dragging
of magnesium atom atmosphere. It is well supported by the fact that the activation
associated with determining the best value of n for experimental creep data, which
span over a very limited range of strain rates (Li and Langdon, 1999b). The
reported fact is in agreement with the earlier work carried out by Cadek and Sustek
(1994), which suggests that the creep data of MMCs should extend over at least
reinforced with 20 vol% of irregular shaped Al2O3 indicates that the true stress
into account (Li and Langdon, 1997a, 1998a). In addition, the true activation
matrix (Li and Langdon, 1997a). The results indicate that the creep in 6061Al
for Al2O3p/7005Al composite, the true stress exponent appears to be ~ 4.4 and the
43
matrix composites is controlled by creep of matrix alloys and the creep behavior of
composites could be classified into following two different classes: (i) class M
solutions (Li and Langdon, 1998b). Similar observations are also reported by Li
and Langdon (1999a) during reanalysis of the available creep data of 20 vol%
It is quite evident from the literature consulted that, though, some of the
researchers (Mishra and Pandey, 1990; Pandey et al, 1992; Gonzalez and Sherby,
1993; Pandey et al, 1994) have used a true stress exponent of 8 to describe the
steady state creep in Al-SiCp,w (subscript ‘p’ for particle and ‘w’ for whisker)
composites but a number of other researchers (Park et al, 1990; Mohamed et al,
1992; Park and Mohamed, 1995; Cadek et al, 1995; Li and Mohamed, 1997; Li
and Langdon, 1997a, 1999a; Yoshioka et al, 1998) have observed that a stress
steady state creep data noticed for discontinuously reinforced Al-SiC composites.
materials may not survive alone. Functionally graded materials (FGMs) are a new
generation of engineered materials that are gaining interest in recent years. The
concept of FGM was first introduced in 1984 in Japan as ultra light temperature-
44
resistant material for space vehicles (Koizumi, 1993). FGMs also find applications
(Noda et al, 1998; Librescu and Song 2005). In FGMs the volume fraction of two
certain dimension(s) of the structure (Suresh and Mortensen, 1998; Reddy, 2000).
FGMs change smoothly and continously from one surface to the other, thus
eliminating the interface problems and also diminishes the concentration of thermal
reinforcement in certain regions (refer Fig. 2.9) where enhanced properties like
increased – modulus, strength and wear resistance are required (Jolly, 1990;
Koizumi, 1995, 1997; Hirai, 1996; Takezono et al, 1996; Akira and Watabane,
1997; Pattnayak et al, 2001; Zhu et al, 2001; Kieback et al, 2003). In addition to
excellent mechanical and thermal properties, the FGMs also possess the advantage
(Noda and Tsuji, 1990; Nagata et al, 1990; Hashida and Takahashi, 1990;
Nakagaki et al, 1991; Erdogan and Wu, 1992; Kumakawa et al, 1992; Teraki et al,
1992; Arai et al, 1993; Ishizuka and Wakashima, 1994; Finot et al, 1994;
Kumakawa et al, 1994; Jin and Noda, 1994; Noda and Jin, 1994; Fukui and
Bowen, 1994; Blumm et al, 1994; Kawasaki and Watanabe, 1994; Pindera et al,
1994; Rabin and Shiota, 1995; Ho and Lavernia, 1996; Tsuda et al, 1996; Noda et
al, 1998). Aerospace industry extensively uses FGMs, as it is desired that the
materials at the surface of space crafts must withstand temperature as high as 2100
K apart from bearing a temperature difference of 1600 K, which may be easily met
Functionally Graded piezoelectric actuators (Li et al, 2003), heated floor systems
(Takeuch et al, 2003), metal/ceramic armor, prosthesis joint with increased adhesive
strength and reduced pain (Quin and Datta, 2004), Thermal Barrier Coatings (TBCs)
for combustion chambers (Ivosevic et al, 2006), thermal protection systems for
spacecraft, hypersonic and supersonic planes (Leushake et al, 2004), rotating blades
in helicopters and turbomachinary (Oh et al, 2005) and smart structures (Ding et al,
2003).
46
chemical vapor deposition (Kim et al, 2005), spark plasma sintering (Shen and
Nygren, 2002; Tokita, 2003) and centrifugal casting (Biesheuvel and Verweij,
2000; Velhinto, 2003). These methods are used to manufacture FGMs with their
The reduction of gas consumption and weight of the car are the motivating
composite, under proper working conditions, is much superior to cast iron, which is
and aluminum alloy matrix, centrifugal separation occurs and higher density
constituent moves towards the outer zones and vice versa. The concentration
profile of ceramic particles, in the radial direction can be controlled and optimized
content and size of ceramic particles, and temperature of molten aluminum. The
47
ceramic particles (Kang and Rohatgi, 1996; Liu et al, 1996; Gao and Wang, 2000).
cylinder liners made of Functionally Graded (FG) Al-SiC and Al-Al2O3 composite.
An attempt was also made to analyze the role of process parameters in centrifugal
liner. The ideal reinforcement distribution was achieved for some combination of
content of reinforcement.
In the past few years, the elastic stresses in FGM subjected to thermo-
mechanical loading have been analyzed by many researchers (Arai et al, 1990; Fukui
and Yamanaka, 1992; Erdogan and Wu, 1993; Hirano and Teraki, 1993; Obata and
Noda, 1994; Tanigawa, 1995; You et al, 2007; Yang, 1998). However, the studies
This approach is based on subdividing the solid model into sub-regions and
48
primary materials available to the Solid Freeform Fabrication (SSF) machine. The
role of design rules restricting the maximum and minimum concentrations has also
Zhu and Miller (1999) examined creep behavior of FGM provided with a
thermal gradient in the FGM was produced by heating the ceramic surface with
laser. The ceramic layer was observed to have primary creep. The time,
gradients in FGM by conducting a number of 3-point bend tests. The study reveals
that the three point bending strength of Al-Al3Ni FGM could adopt the two
decreases with the increase in volume fraction of Al-Al3Ni. The strength of Al-
Al3Ni FGM depends on the cleavage fracture strength of Al3Ni and obeys the law
σ FGM = (1 − f ) σ Al + fσ Al Ni
3
(2.22)
20, 40, 60 and 80 wt% of Si3N4 with corresponding decrease in MoSi2 content.
49
Each layer was 2 mm thick and possessed uniform distribution of Si3N4. The study
indicates that the creep rates in a single layer decrease with increasing content of
Si3N4.
experiment was implemented for the creep test numerically and the creep
parameters were estimated. The numerical results indicate that the creep
phenomenon is obvious not only for the metal-rich interlayers but also for the
larger than that observed for pure metal under the same load when modulus of the
Rotating disk provides an area of research and studies due to their vast
utilization in rotating machinery viz. steam and gas turbine rotors, turbo generators,
computer disk drives (Gupta et al, 2005; You et al, 2007; Hojjati and Hassani,
2008). In most of these applications, the disk has to operate at elevated temperature
high speed (Farshi and Bidabadi, 2008). As a result of severe mechanical and
deformations, thereby affecting its performance (Laskaj et al, 1999; Farshi et al,
2004; Gupta et al, 2005). For example, in turbine rotor there is always a possibility
50
that heat from the external surface is transmitted to the shaft and then to the
bearings, which may adversely affect the functioning and efficiency of the rotor
Optimal and more reliable design of rotating disks has long been an
disk and physical properties of the disk material, optimal and more reliable design
of rotating disk for given operating conditions (i.e. load, speed, operating
process are the geometrical parameters like the mean radius and thickness of the
disk, and material properties such as density, elastic modulus and Poisson’s ratio
disk can be found in most of the standard text books and literature (Malkin, 1934;
Finnie and Heller, 1959; Lubhan and Felger, 1961; Odqvist, 1974; Findely et al,
1976; Kraus, 1980; Boyle and Spance, 1983; Skrzypek and Hetnarski, 1993;
Nabarro and Villiers, 1995; Penny and Mariott, 1995). Timoshenko and Goodier
(1970) was the first to obtain closed form solutions for rotating homogeneous disks
but without considering temperature gradient. Reddy and Srinath (1974) and Chang
of density gradient in the disk significantly affects the distribution of stresses and
displacements. Zhou and Ogata (2002) obtained closed form solutions for rotating
51
solid disk made of cubic anisotropic material by using direct displacement method.
The displacement, strain and stress were expressed as a simple function of polar
coordinates. Orcan and Eraslan (2002) obtained analytical solution for elastic-
function. The analysis assumed the Tresca’s yield criterion and its associated flow
rule, and linear strain hardening material behavior. The solution obtained was
verified by comparing it with the solution available for uniform thickness disk. It is
observed that with the reduction in disk thickness the plastic limit angular velocity
optimizing the profile of rotating disk (Fox, 1970; Zienkiewicz and Campbell,
1973; Malkov and Salganskaya, 1976; Pederson, 1981; Wang and Gallagher, 1985;
simultaneously subjected to mechanical and thermal loads. The disk may also be
subjected to internal pressure due to shrink fitting on a shaft. In addition, the blade
effects may also be modeled by applying external tensile load at the outer radius of
the disk. When the disk rotates at significant angular velocity, while the gases cross
Kollman (1978, 1981, 1984) solved the problem of shrink fitted disk, both
rotating as well as non-rotating, using Tresca’s yield criterion. Yeh and Han (1994)
rotating disk with arbitrary thickness and operating under thermal loading.
52
Gas turbine disks mostly operates under high temperature gradients and is
also subjected to high angular velocity. High speed results in large centrifugal
forces in the disk and simultaneous presence of high temperature reduces the
the disk. In order to attain accurate and reliable analysis of stress distribution in the
disk, the solution should consider changes in the material properties caused by
inhomogeneous disk model with variable thickness. Using the variable material
properties method (Jahed and Dubey, 1997; Jahed and Sherkatti, 2000; Jahed and
Shirazi, 2001), stresses were obtained in a rotating disk operating under steady
function was the total weight of the disk and the constraints were imposed on the
stresses, which were kept less than the yield stress of the material. The disk profile
was optimized and the final solutions were obtained. The study indicates that the
solutions of optimization process for different initial profiles of the disk with
similar specifications are unique when the inscribed hypersphere radii in the last
solution stages are equal. Results obtained were compared with the published data
Farshi and Bidabadi (2008) pointed that the rotating disks are subjected to
secondary creep effects during most of their useful lives, thus it is important that
they should be optimized for minimum weight while considering only the steady-
state creep stresses. They assumed variable physical properties of the disk material,
which was assumed to operate under a high temperature gradient. The study
53
proposed a procedure for weight minimization during steady state creep. The
method estimates the disk thickness profile by keeping its weight to be minimum
while the equivalent secondary creep stresses in the rotating disk at all points
analyze stresses and strains in rotating disks having nonuniform thickness and
the numerical solution was based on the solution of governing differential equation
using Runge–Kutta’s method for elastic and plastic regimes. Finite element
modeling of the problem was also carried out by using commercially available
software. The results obtained by these three methods generally show good
agreement. The study also reveals that the VMP method is a reliable means for the
circumferential stress components are higher than the radial stress components. The
flow are highest at the inner surface. At all angular velocities, the radial
displacements for both elastic and plastic solutions have higher values at the inner
continuously from point to point, may be considered in order to make optimal use
control of stresses and deformations along with controlling the undesirable features
such as internal residual stress (Nagatha and Takahashi, 1995; Williamson et al,
modulus leads to equal radial and circumferential stresses in the rotating disk.
and strains in rotating axisymmetric components such as disks and rotors. The
Durodola and Attia (2000) carried out similar analysis for rotating hollow and solid
disks subjected to centrifugal body load. The disks were assumed to be made of FG
long fibers. Stresses and displacements in the disks were obtained by direct
solution for hollow and solid rotating axisymmetric disk made of FGM under plane
stress condition.
55
bending and steady-state thermal loading. The material properties of the disk were
an exact solution for displacement field was obtained. It is observed that for
disks having variable thickness profile. The material properties and disk thickness
effects of material grading index and geometry of the disk were investigated on the
stresses and displacements in the disk. It is observed that the maximum radial stress
in the solid FG disk with parabolic thickness profile is not at the centre like
uniform thickness disk. The study reveals that FG disk with parabolic concave or
hyperbolic convergent thickness profile can be more efficient than the uniform
thickness disk.
Sharma and Sahni (2009) used Seth’s transition theory to obtain elastic-
plastic stresses in thin rotating disks made of transversely isotropic and purely
elasticity solution for FG hollow and solid rotating disks. The study aimed to
dimensional one. It is revealed that for thin disks the two-dimensional solution
Sharma and Sahni (2011) used transition theory to obtain elastic – plastic
observed that the rotating disk made of incompressible material with inclusion
requires higher angular speed to yield at the internal surface as compared to disk
made of compressible material. For disk with exponentially varying thickness, high
angular speed is required for initial yielding at the internal surface as compared to
flat disk. It is also concluded that the disk made of isotropic compressible material
Disks of gas turbines, jet engines, and automotive and aerospace braking
systems usually operate at relatively higher angular speed and high temperature or
important for these applications. Most of the published work on creep in rotating
disk is dedicated to the steady state creep behavior since a major part of the
distributions in a rotating forged disk made of 12% chromium steel at 1000 0F. The
carried out using von Mises and Tresca yield criteria, while the creep behavior was
described by power law. The stress distributions obtained using Tresca and Mises
criteria do not differ significantly. The normalized strain curves calculated for
various values of stress exponent (n) corresponding to both the yield criteria are
observed to practically coincide for n = 6 to 9. The study also reveals that the
theoretical and experimental stresses are in better agreement if the Tresca criterion
is used. The creep deformations estimated using Mises theory are found to be quite
creep at elevated temperature. The formulas were based on the Tresca’s yield
criterion and the associated flow rule and give reasonable results when compared
with the available experimental data (Wahl et al, 1954). The method proposed was
also applied to calculate the transient change in stress when the stress distribution
changes from an initial to a steady state condition during the starting period.
Wahl (1957) utilized the formulas derived in their previous work (Wahl et
al, 1954) to construct the design charts of stress distribution in constant thickness
disks, undergoing steady state creep, for different values of stress exponent (n) and
diameter ratios. In all the cases, the disks were subjected to a radial peripheral load
58
to simulate the effect of blade loading. The steady state creep rate was expressed as
Wahl (1958) extended his previous work (Wahl, 1957) on rotating disks
having central holes and undergoing steady state creep to cases where the radial
and tangential stresses are equal over all or a portion of the disk. Both constant and
variable thickness disks were considered and the charts were presented for
determining the ratios of peak stress to average stress for various diameter ratios,
disk contours and stress exponent. The study indicates that the variable thickness
disk has somewhat lower ratios of peak to average stress than those observed for
stress theory (Tresca yield criterion) associated with the Mises flow rule and the
assumption that the tangential stresses in the disk are invariably greater than the
radial stresses, except at the center of the disk. The results obtained using Mises
Ma (1960) extended his work for variable thickness solid disks, used in gas
turbine and jet engine, operating at uniform temperature. The study used Tresca’s
criterion and its associated flow rule while the steady state condition was described
by exponential creep law. It is revealed that the stress distributions over central
59
portion of variable thickness disk are quite different from those observed in a
Ma (1961; 1964) further extended his analysis for variable thickness disk
operating under variable temperature. The steady state creep was described by
either exponential creep law (Ma, 1961) or power law (Ma, 1964). The study
reveals that the proposed analyses can be used to obtain closed form solutions for
complex disk design problem with great simplicity instead of using tedious
numerical solutions.
Wahl (1963) investigated the effects of initial transient period on the long-
time creep tests of rotating disks by using both time-hardening and strain-hardening
relations. The results obtained were applied to the long-time spin-tests conducted
on steel disks at 1000 0F (Wahl et al, 1954). The study reveals that by considering
the effects of transient period, there is no appreciable impact on the over-all creep
deformation noticed during the spin tests. However, when the creep deformations
Arya and Bhatnagar (1979) analyzed the creep stresses and deformations in
function of time. A numerical example was worked out to investigate the effect of
distributions in the disk. It is observed that the tangential stress at any radius and
the tangential strain at the inner radius of the disk decrease at all times for an
anisotropic material. The time taken to reach steady state distribution decreases
disks having constant and variable (linear and hyperbolic) thickness. The creep
stresses and strains were obtained for different cases of anisotropy by using
Norton’s power law creep model. The study reveals that the selection of a certain
type of material anisotropy and an optimum disk profile would lead to a better disk
design.
and creep rates in a thin rotating disk having variable thickness and variable
density. The study reveals that a rotating disk with radially varying density and
thickness ratio is on the safer side of design as compared to a flat disk having
variable density.
operating temperature on the steady state creep behavior of a rotating disk made of
parameters appearing in the creep law were extracted from the available
experimental creep results reported by Pandey et al (1992) for Al-SiCp. The study
reveals that particle size, particle content and operating temperature do not have
significant effect on the distribution of radial stress but their effect on the
distribution of tangential stress is sizable. Both the radial and tangential strain rates
in the disk decrease by several orders of magnitude with decreasing particle size,
(2004a) also made similar observations, however they used Sherby’s law to
describe the steady state creep behavior of the disk material (Al-SiCp). The study
61
also reveals that for a given operating condition, the strain rates in rotating disks
can be controlled by selecting optimum particle content and/or particle size of the
reinforcement.
Singh and Ray (2001; 2003a) analysed steady state creep in a rotating disk
made of isotropic FGM (Al-SiCp) and operating at a constant temperature. The disk
was assumed to have linearly decreasing content of SiCp from the inner to outer
radius and creeping according to Norton’s power law. It is observed that the steady
state creep response of the FGM disk in terms of strain rates is significantly
Singh and Ray (2002) studied the influence of anisotropy, induced due to
processing, on the steady state creep in a rotating disk made of 6061Al-20 wt%
SiCw and operating at 561 K. The creep behavior of the disk material was described
by Norton’s power law. It is observed that the presence of anisotropy in the disk
leads to significant reduction in both the tangential and radial strain rates over the
loading and temperature gradient. The method uses the basic solution for a rotating
disk made of uniform isotropic material and generates the solution for disk made of
and the results obtained were compared with those estimated by FEM technique. In
Singh and Ray (2003b) analyzed steady state creep in a rotating disk made
stress. The study used Norton’s power law and the newly proposed yield criterion,
which at appropriate limits reduces to Hill anisotropic and Hoffman isotropic yield
anisotropic disk but without residual stress. The presence of tensile residual stress
in the disk leads to significant increase in creep rate as compared to that observed
stress, the radial strain rate becomes tensile in the middle of the disk, however, it
remains compressive towards the inner and outer radii. Similar results were also
noticed in the subsequent study by Singh and Ray (2004), based on the effect of
thermal residual stress in a rotating disk made of isotropic 6061Al-20 vol% SiCw
and undergoing steady state creep. The study used Hoffman yield criterion for
isotropic material.
operating in the presence of radial thermal gradient. The creep behavior of the
composite was described by Sherby’s law. The study indicates that the steady-state
strain rates in the FGM disk are significantly lower than that observed in an
isotropic disk having uniform distribution of SiCp, when both the disks operate
under thermal gradient. The study also reveals that the strain rates in composite
63
disk operating under thermal gradient are lower as compared to a similar disk
gradient.
Gupta et al (2007) used Artificial Neural Network (ANN) for predicting the
temperature. The analysis of steady state creep, described by Sherby’s law, was
carried out for various combinations of SiCp size and content, and operating
temperature. The creep parameters were extracted from the limited experimental
uniaxial creep data available for Al-SiCp. The results obtained were used to train
the ANN model based on back propagation learning algorithm with SiCp size,
SiCp content and temperature as the input and stresses and strain rates as the output
parameters. The predictions obtained from the ANN model were compared with
observed between the creep stresses and strain rates predicted by ANN model and
estimated analytically.
Singh (2008) used Norton’s creep law to analyze steady state creep in a
rotating disk made of 6061Al-20 vol% SiCw with varying extent of anisotropy,
remains compressive for isotropic composite disk (α = 1.0) and anisotropic disk
having α > 1, becomes tensile in the middle of the disk when the extent of
anisotropy parameter α < 1. By changing the extent of anisotropy from α > 1 to α <
1, the variation of tangential strain rate in the disk remains similar, however, its
introduces significant change in the strain rates, though its effect on the resulting
6061Al matrix composite. The model was used to investigate the effect of SiC
morphology on the creep behavior of composite disk. The steady state creep
behavior was described by Sherby’s creep law. The study reveals that the creep
stresses and creep rates in the disk are significantly affected by the morphology of
SiC. The steady state creep rates in a whisker reinforced disk are observed to be
Singh and Rattan (2010) analyzed steady state creep in a rotating disk made
obtained were compared with those using von Mises yield criterion. It is observed
that the distribution of stress in the disk is not too much affected in the presence of
phase specific thermal residual stress. The presence of residual stress leads to
increase the tangential strain rate, particularly in the region near the outer radius of
the disk, as compared to that observed in a similar disk but without residual stress.
The radial strain rate, which is compressive, changes significantly in the presence
of residual stress and even becomes tensile in the middle of the disk.
state creep behavior in rotating disks made of isotropic aluminum matrix composite
containing linear and quadratic distributions of SiCp in the radial direction. The
disks were assumed to operate under a radial thermal gradient, originating due to
65
braking action as estimated by FEM analysis. The steady state creep behavior of
the disks was described by Sherby's law. Based on the developed model, the
distributions of stresses and strain rates were obtained and compared for various
FGM disks containing the same average amount (20 vol%) of SiCp. The study
reveals that the creep stresses and steady state creep rates in a rotating FGM disk
not have considerable influence on stresses. However, the minimum and the most
to FGM disk made of aluminum with 0% SiC at the inner surface and 40% SiC at
the outer surface. It has also been found that the stresses, displacement and creep
strains change with time at a decreasing rate in such a manner that after almost
elastic stresses and deformations in rotating disk made of monolithic material has
leads to consistently higher value of activation energies for creep than anticipated
assuming the flow to be controlled by creep of matrix material so that the true
stress exponent (n) is ~3 or ~5 (Cadek et al, 1995; Li and Langdon, 1999). In this
The stresses produced in the disk are due to rotary motion, which can be
analysis of the steady state creep in rotating disks made of composite material
loadings, studies have been conducted to analyze creep in rotating disk made of
isotropic FG composites (Singh and Ray; 2001; 2003a Gupta et al, 2004b;
2005). In all these analyses, the disk was assumed to be of constant thickness
Keeping this in view, it may be interesting to extend the above analysis for
which are responsible for different yield strengths of the composite under
tension and compression. Singh and Ray (2003b) investigated steady state
investigate the effect of thermal residual stress on the creep behavior of rotating
pertaining to creep behavior of rotating composite disk, reveals that the effect of
transient period has not been considered while evaluating creep response of the
composite disk. Neglecting the transient phase can lead to errors in estimating
the creep deformation of the disk, whose magnitude may be of the order of
investigate the effects of initial transient period on the long-time creep behavior
2.7 OBJECTIVES
On the basis of literature consulted and the facts reported in the previous
section, the following objectives are set for the present study.
ii. To analyze the steady state creep in a rotating composite disk having
thickness profile.
2.8 METHODOLOGY
In order to meet the objectives outlined in the proposed research work, the
to disk material.
69
(ii) The equilibrium equation of rotating disk will be solved along with
disk.
(iii) The material constants appearing in the creep law and yield criteria
analysis.
Fig. 2.3: Simplified deformation mechanism map (Ashby and Frost, 1975)
73
Fig. 2.5: Analysis of the competing processes determining the creep curve (Andrade, 1957)
75
Fig. 2.6: Creep strain rate as function of total strain (Deiter, 1988)
76
Fig. 2.8: Variation of steady state creep rate with applied stress for PM 15 vol
% SiCp/8009Al composite (Ma and Tjong, 2000)
78
THot
Ceramic Phase
Transition region
`
Metallic Phase
TCold
Metallic matrix
Ceramic inclusion
Ceramic phase
Inclusion phase 1
Inclusion phase 2