Introduction To Creep Mechanics PDF
Introduction To Creep Mechanics PDF
Introduction To Creep Mechanics PDF
net/publication/336510171
CITATIONS READ
0 1
2 authors:
Some of the authors of this publication are also working on these related projects:
Free vibrations and buckling of laminated composite shells with low reduced shear modulus: effect of shears View project
All content following this page was uploaded by Johanna Eisenträger on 14 October 2019.
Synonyms
Definitions
1
2 Holm Altenbach and Johanna Eisenträger
take place under constant volume and the superimposed hydrostatic pressure does
not influence creep.
In 1934, multi-axial creep equations were suggested by F. K. G. Odqvist (see
Odqvist, 1974, among others)
ε̇ε = f σeq s
with the strain rate tensor ε̇ε and the equivalent stress in the von Mises sense
2 3
σeq = s ··ss.
2
Here, s ··ss denotes the double scalar product of the stress deviator s . For the confor-
mity with Norton’s law, the following expression can be used for the scalar func-
tion f :
3 n−1
f (σeq ) = Aσeq .
2
Odqvist’s theory is similar to the von Mises theory for time-independent plasticity.
The approach is straightforward to formulate since one only needs two parameters
obtained through a one-dimensional creep test and it can be extended to primary and
tertiary creep by introducing hardening, softening, and damage parameters (see, for
example Naumenko and Altenbach, 2016). The principal idea of the extension of
the secondary
4 creep equations to primary and Holm
tertiary creepandisJohanna
Altenbach demonstrated
Eisenträger in Fig.
1.1. In a first step, the equation for secondary creep will be established for the range
σ = const, T = const
Fracture
ω = ω∗
Creep strain
σ2 > σ1 σ1 Tertiary
ω̇ = gω (σ , H, ω , T)
Secondary
cr = g (σ , T )
ε̇min ε
Primary
Ḣ = gH (σ , H, T )
Time
cr cr = 0 Constitutive equation,
ε̇ = gε (σ , H, ω, T ), εt=0
Ḣ = gH (σ , H, ω, T ), Ht=0 = 0 Evolution equation (hardening),
ω̇ = gω (σ , H, ω, T ), ωt=0 = 0 Evolution equation (damage)
Fig. 1.1 Classical phenomenological modeling of creep with a hardening variable H and an
ε̇ cr = gωε ((σ
isotropic damage variable σ , H,
=ωconst,
cr = 0 Constitutive equation
, T), εT
t=0= const)
Ḣ = gH (σ , H, ω , T), Ht=0 = 0 Evolution equation (hardening)
ω̇ = gω (σ , H, ω , T), ωt=0 = 0 Evolution equation (damage)
Fig. 1 Classical phenomenological modeling of creep with hardening H and isotropic damage ω
(σ = const, T = const)
• Sintering
They are related to different mechanisms which depends on the temperature and
stress levels. Such mechanisms are bulk diffusion, climb (the strains are accom-
plished by a climb), climb-assisted glide (dislocations can go around obstacles),
grain boundary diffusion or thermally activated glide. Some mechanisms of creep
have not been verified by direct microstructural examination yet.
In dependence of the mechanism various equations are suggested for the descrip-
tion of creep. Not all are suitable for creep mechanics analyses. Below some of these
4 Holm Altenbach and Johanna Eisenträger
of the minimum creep rate. Afterwards, the evolution of the hardening variable can
be established. Last but not least, the damage evolution will be introduced to de-
scribe the tertiary creep range. In the end, one obtains a coupled system of ordinary
differential equations with respect to time.
Probably the first approach to describe primary creep was presented in 1938 by
A. Nadai, who suggested an equation of state for strain hardening (Nádai, 1963). In
addition, a different equation was introduced for time hardening by Finnie (1960).
First theories for tertiary creep were developed by Kachanov (1958) and Rabot-
nov (1959), who introduced the concept of material deterioration. In their simplest
form, the theories are based on two material parameters derived from a standard
creep rupture curve, which illustrates the dependence of the time to rupture on the
applied stress level. Further information can be found for example in Kowalewski
and Ustrzycka (2018).
In addition, Knych et al (2018) report on current trends in creep mechanics. Other
types of time-dependent material behavior, such as viscoelasticity and viscoplastic-
ity, are presented in Milewski (2018); Sumelka and Łodygowski (2018). Special
attention to solving problems is paid in Naumenko and Altenbach (2007, 2016,
2019); Rabotnov (1969); Szuwalski and Ustrzycka (2018) and the references within.
Furthermore, information on the micromechanics of creep is given in Kowalczyk-
Gajewska (2018).
Based on the terminology of da Costa Andrade (1910, 1914), one distinguishes three
stages, while the elastic range is usually ignored. In the initial stage (primary or de-
celerate creep, sometimes also transient creep), the strain rate is relatively high, but
decreases with increasing time and strain due to a process analogous to work hard-
ening at lower temperatures. For instance, the dislocation density increases and in
many materials a dislocation subgrain structure is formed and the cell size decreases
with strain (Courtney, 2005).
With increasing deformation, the strain rate will attain its minimum and become
nearly constant as the secondary stage begins. This is due to the balance between
work hardening and annealing (thermal softening). The secondary stage is also of-
ten referred to as steady-state creep. However, it is worth pointing out that various
microstructural processes take place during this stage. Now, recovery effects are
concurrent with deformation. Note that changes in material strength are only minor
during these first two stages of creep.
The last stage (tertiary or accelerated creep) is again related to an increase of
the strain rate. Here, softening processes are dominant. At the same time distributed
damage occurs and finally fracture can be observed.
1 Introduction to Creep Mechanics 5
In material science, one distinguishes the following types of creep (Cotrell, 1967):
• dislocation creep,
• Nabarro-Herring creep
• Coble creep
• solute drag creep
• dislocation climb-glide creep
• Harper-Dorn creep
• sintering
They are related to different microstructural mechanisms, which depend on the tem-
perature and stress levels. Such mechanisms are for example bulk diffusion, climb
(the strains are accomplished by climbing dislocations), climb-assisted glide (dis-
locations can go around obstacles), grain boundary diffusion or thermally activated
glide. Note that some mechanisms of creep have not been verified by direct mi-
crostructural examinations yet. The so-called deformation-mechanism maps sum-
marize the relations between deformation and mechanisms for many materials, cf.
Frost and Ashby (1982); Nabarro and de Villiers (1995); François et al (2012).
Deformation-mechanism maps represent the dominant deformation mechanisms in
diagrams based on the normalized shear stress, which is displayed with respect to
the homologous temperature.
Depending on the microstructural mechanisms, various equations are suggested
for the description of creep. However, not all are suitable for structural analysis in
creep mechanics. Below, some of these equations are presented.
The general creep equation is formulated for the creep strain rate
Q
Cσ m −
ε̇ = e kT , (1.1)
db
whereas C is a parameter depending on the material and the particular creep mech-
anism, m and b are exponents related to the creep mechanism, Q is the activation
energy of the creep mechanism, d denotes the average grain size of the material,
k ≈ 8.31696 J (mol K)−1 is Boltzmann’s constant, and T is the absolute tempera-
ture. Note that Equation (1.1) is a power law function.
6 Holm Altenbach and Johanna Eisenträger
At high stresses (relative to the shear modulus G), creep is controlled by the move-
ment of dislocations. For dislocation creep, the activation energy Q is related to
self-diffusion, the exponent m is in the range 4 . . . 6, and b = 0 holds. It is obvious
that dislocation creep is strongly influenced by the applied stress and the intrinsic
activation energy, while the influence of the grain size can be neglected. Note that
some alloys feature a very large stress exponent (m > 10). In this case, a threshold
stress σth will be introduce: creep cannot be measured below this value. Equation
(1.1) should be modified as follows
Q
A(σ − σth )m −
ε̇ = e kT , (1.2)
db
where A, Q and m can all be related to conventional mechanisms. The applied stress
tends to drive the dislocations past the barrier. After bypassing an obstacle, dis-
locations get into a lower energy state. A significant part of the work required to
overcome the energy barrier of passing an obstacle is provided by the applied stress,
while the remaining impulse is provided by thermal energy. As a result, creep in-
creases in these cases.
In contrast to dislocation glide creep, which does not involve atomic diffusion,
Nabarro-Herring creep belongs to the diffusional creep mechanisms. This creep
mechanism dominates the material behavior particularly at high temperatures and
low stresses. The sides of the crystal are subjected both to tensile and compressive
stresses and the atomic volume is changed due to the applied stresses: it increases
under tension and decreases under compression. The activation energy for vacancy
formation is changed by ±σ Ω , where Ω represents the atomic volume, the sign +
refers to compressive regions and a negative sign describes tensile regions. Since
the fractional vacancy concentration is proportional to exp(− Qf ±σkT ), where Qf is
Ω
Coble creep is another form of diffusional creep. The atoms diffuse along grain
boundaries to elongate the grains in the stress direction. In this case, the influence
of the grain size on creep deformations is significantly higher compared to Nabarro-
Herring creep. Thus, this mechanism is of higher importance for materials made
of very fine grains. Note that the Boltzmann’s constant k is related to the diffusion
coefficient of atoms along the grain boundary, Q describes the grain boundary dif-
fusion, and m = 1 and b = 3 hold for the exponents. Since the activation energy
for grain boundary diffusion is smaller than in the case of self-diffusion, this creep
mechanism occurs at lower temperatures in comparison to Nabarro-Herring creep.
Although Coble creep is temperature-dependent as well, the influence of tempera-
ture is not as strong as for Nabarro-Herring creep. It also exhibits the same linear
dependence on stress as Nabarro-Herring creep. In general, the diffusional creep
rate should be obtained as a sum of the Nabarro-Herring creep rate and the Coble
creep rate. Note that diffusion creep results in grain-boundary separation and the
occurrence of voids or cracks between grains. As a result of these mechanisms,
grain-boundary sliding starts. If grain-boundary sliding cannot accommodate the
incompatibility, the voids at the grain boundaries lead to the initiation of creep frac-
ture.
Creep failure must be prevented, and the lifetime of components should be pre-
dicted in an accurate manner. Consequently, there is a high demand for constitutive
models accounting for not only one-dimensional effects, but also three-dimensional
stress and strain states. As additional challenges, non-isothermal, and cyclic operat-
ing conditions should be taken into account. Thus, not only long-term creep strains
should be predicted accurately, but also short-term plastic strains as well as the in-
teraction of both deformation modes. Constitutive equations for metallic materials
are well presented in various monographs and textbooks, among them Hult (1966);
Odqvist (1974); Penny and Marriott (1995); Naumenko and Altenbach (2007); Bet-
ten (2008); Naumenko and Altenbach (2016).
An essential part of creep models is the proper description of the stress and tem-
perature influence by so-called response functions. These functions are specific for
each deformation mechanism, and the choice of these functions can be based on
information from deformation mechanism maps (Frost and Ashby, 1982; Nabarro
and de Villiers, 1995; François et al, 2012). In the following, let us assume that
8 Holm Altenbach and Johanna Eisenträger
the influence of the stress and temperature can be separated. We will introduce
some commonly used stress and temperature response functions with respect to one-
dimensional stress and strain states.
In the simplest case, the creep strain rate is a function of the stress σ and the
temperature T :
ε̇ = f (σ , T ). (1.3)
This function can be extended to include further influences, but in this case the effort
for the identification of material parameters increases in accordance to the number
of variables. Even in the simplest case, the identification is not straightforward since
all influences are usually coupled.
A separation ansatz can simplify this procedure. To define the dependence of the
creep strain rate ε̇ on the stress σ and the temperature T , the following approxima-
tion is frequently introduced:
ε̇ = fσ (σ ) fT (T ), (1.4)
which is a product of the stress response function fσ and the temperature response
function fT .
The Norton power law represents one of the most commonly used expression for
the stress response function:
fσ (σ ) = Aσ n . (1.5)
The material parameters A and n (creep exponent) should be estimated from creep
experiments. Diffusion creep is usually described by employing a linear stress func-
tion (Herring, 1950; Harper and Dorn, 1957; Coble, 1963; Lifshitz, 1963), i.e. n = 1:
fσ (σ ) = Bσ , (1.6)
with the material parameter B. It is obvious that such a behavior is similar to the
behavior of a viscous fluid. Another common choice to describe creep at low tem-
peratures and with respect to wide stress ranges is a hyperbolic function (Dyson and
McLean, 2001):
fσ (σ ) = C sinh (Dσ ) . (1.7)
Again, the material parameters C and D must be determined based on experimental
data.
To express the dependence of the creep strain rate on temperature, the Arrhenius
function is a popular choice (Dorn, 1955; Ilschner, 1973):
Q
fT (T ) = α exp − (1.8)
kT
with the material parameter α, the activation energy Q, and the Boltzmann constant
k. As has already been pointed out, the temperature exerts a significant influence on
creep and relaxation phenomena. This is due to the fact that creep is based on mi-
crostructural processes, which are highly dependent on the temperature. However,
because of the great variety of materials affected by creep, the classification of creep
1 Introduction to Creep Mechanics 9
with respect to temperature is not uniform in literature, i.e. a clear and widespread
definition of low and high-temperature creep does not exist. Note that temperature
ranges in creep are often indicated with respect to the liquidus temperature TL of
the material under consideration. Other indicators are the solidus TS or melting Tm
temperature.
According to Naumenko and Altenbach (2016), high temperature materials are
used at a temperature range of 0.3–0.7 Tm . In Frost and Ashby (1982), it is stated that
polycrystalline solids start to creep at 0.5 TL , whereas a temperature level of 0.9 TL
is referred to as “very high” temperature level. In addition, according to McLean
(1966), the temperature range 0.3–0.9 TL is most important for engineering applica-
tions, while in this main creep range, two separate creep regimes are distinguished,
low temperature (LT) creep at 0.3–0.5 TL and high temperature (HT) creep at 0.5–
0.9 TL . Higher temperatures are related to other mechanisms and should be presented
by the so-called power-break law.
There are also other approaches (Eisenträger and Altenbach, 2020). However,
practical applications involving structural analyses still represent a major challenge.
Only approaches using advanced rheological models, which are highly non+linear,
but reflect the mechanisms separately, simulate real creep behavior with an adequate
accuracy and involving only moderate computational effort, cf. e.g. Eisenträger et al
(2018a,b,c); Eisenträger (2018).
The equations suggested in the previous section can be extended to multi-axial stress
and strain states considering the following three main steps:
• substitution of the scalar stress by a tensorial variable,
• substitution of the scalar strain by a tensorial variable, and
• introduction of an equivalent stress.
It seems that the extension is straightforward, but there are several difficulties. It
seems natural to use the stress tensor and the strain tensor. However, with respect to
Bailey’s statements (Bailey, 1935), several questions arise: should one choose the
stress or strain tensor or deviator, respectively. Here, one should also consider the
axioms of rheology (Reiner, 1960) and statements of Palmow (1984).
The last item in above list is much more complicated. Even in the case of
isotropic materials, the choice of an equivalent stress cannot be based on rational
approaches such as balance equations. Instead, the introduction of some hypotheses
and experimental proofs allow the introduction of an adequate expression for the
equivalent stress (for further reading: Altenbach and Öchsner, 2014).
10 Holm Altenbach and Johanna Eisenträger
1.4 Cross-References
References