Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Creep Behavior of Materials: A can be found in Ashby (1999).

Some examples for the


purpose of illustration are listed in Fig. 1. We have
Comparison chosen design requirements associated with either
stiffness or strength, aiming at minimizing the mass.
Design of components required to work at high Note that in this form, one considers the instantaneous
temperature often has to address the requirement of response to a load, and any delayed responses, such as
the materials to have a good ‘‘creep resistance.’’ those expected at high temperatures, are neglected.
Several articles in this encyclopedia deal with this There are several ways of improving on this situation,
requirement for designing components such as pipes, and to extend these methods to compare materials for
boilers, and steam turbines. In these situations, the creep design, at least in the first stages of design.
materials are already ‘‘almost chosen,’’ and the re-
quirements are to improve the resistance to creep, and
prevent the deterioration of materials properties dur- 2. Temperature Dependence of the Various
ing service. Another possibility to improve creep Mechanical Properties
design is to rely on engineering solutions such as The very first limitation of these methods is of course
thermal barriers or local cooling devices to decrease that the properties themselves are temperature de-
the effective temperature that the material experi- pendant. In principle, as far as the instantaneous
ences lower than the operating temperature of the response of the component is concerned, they can be
component itself. Both these solutions, improving readily extended to high-temperature design, provided
the microstructure of the materials or improving the mechanical properties are suitably corrected for
the design of the component are efficient, pro- temperature effects. In order to do so, one has to
vided the proper choice of the materials class has distinguish, at least qualitatively, between several
been done in the first place. classes of properties. Some properties are mildly
The aim of this article is to present methods temperature dependant (such as the density, the
available at earlier stages of design, when the choice of coefficient of thermal expansion, and the specific heat):
the material is still a very open question. The tools their variations with temperature between room tem-
presented here are, in principle, some that could be perature and the melting point are just of a few percent
used to compare materials within a given class, and and, at the level of precision we are concerned with,
from different classes. We will therefore insist on the they may as well be considered as constant. Some
format to be given to data for high-temperature other properties vary significantly with temperature,
resistance of materials so that they can be used for say by a factor of two between room temperature and
comparison purposes. These methods parallel the ones the melting point; for instance, this is the case for
developed for room temperature design, but they elastic moduli, thermal conductivity, and resistivity.
account for some of the specificity of high-temperature For these properties, however, a linear approximation
behavior. We will first outline the performance index for their variation is accurate enough, and thus their
method since it is the basis of this article; we will then values for design purposes can be safely approximated
outline its limitations as far as creep design is con- from existing data in publications such as Waterman
cerned. The various ways of dealing with these and Ashby (1996) for all materials classes, ‘Smithells
limitations will then be presented, dealing first with Metals Handbook’ (1992) for metals, ‘ASM Engin-
limiting operation temperature of materials, then with eered Materials’ (1991) for ceramics, and ‘Saechtling
design for limited time in a steady-state creep regime, International Plastics’ (1983) for plastics. It is worth
and then extending these methods to transient creep noting at this point the special situation of polymers:
regimes. although plastics are usually not suitable for high-
temperature applications, a creep problem may occur
with them at relatively low temperatures. The ref-
1. Performance Index Method for Room
erence temperature for polymers is not the melting
Temperature Design
temperature, but the glass transition temperature.
Consider for instance a rotating fan operating at room Above this temperature, their elastic properties de-
temperature. In the performance index method, a set crease drastically by orders of magnitude, and can
of requirements is structured in terms of constraints certainly not be extrapolated linearly. Finally, another
and objectives. For instance, a component in a rotating class of properties has a very large degradation with
engine should be designed such that it does not yield temperature, often more than one order of magnitude
under the centrifugal load (constraint), and that it has between room temperature and melting point, among
a minimum mass and a maximum safe rotating speed them the strength, the toughness, the oxidation rate.
(objectives). For such a set of requirements, the Most of the strength decreases with temperature
appropriate performance indices to be maximized in (although many intermetallics and related alloys show
an efficient design are σys\ρ, where σys is the yield stress the so-called anomalous behavior, an increase of the
and ρ the density of the material. The method has been yield strength up to a peak value, which makes them
widely developed and a range of performance indices interesting for high-temperature applications). But the

1
Creep Behavior of Materials: A Comparison

Figure 1
Performance indices for lightweight design.

degradation is certainly nonlinear over the whole little risk of substantial creep below one-third of the
range between room temperature and melting tem- melting temperature. This statement may be modified
perature. The commonly observed behavior is a slow in the case of alloys that are hardened by precipitation:
decrease followed by a sharp drop. This allows one to a key temperature is then the solvus temperature of the
define approximately a maximum service temperature hardening phase. For polymers, it is the glass tran-
at which this sharp drop occurs. Below this maximum sition temperature that plays the role of a reference
service temperature, one can safely approximate the temperature, but the manner according to which
decrease in properties by simple laws. Above this properties are evolving may be very different for a
transition temperature, not only do the properties polymer-based composite. The maximum serŠice tem-
themselves evolve more rapidly, but time-dependant perature can be seen as a first attempt to deal with the
behavior becomes non-negligible, in particular, creep. limitations stated in Sect. 1: well below this tem-
perature, the performance indices such as the ones
given in Fig. 1 will still be useful provided the
properties are corrected for their temperature de-
3. Maximal Use Temperature for Materials
pendence. This correction might not be very important
High temperature is not a straightforward concept, for stiffness design, it is, however, certainly crucial for
but depends very much on the materials under strength design.
consideration. For pure metals and ceramics, there is Figure 2 shows a ‘‘selection map,’’ drawn using CES

2
Creep Behavior of Materials: A Comparison

Figure 2
Maximum service temperature as a function of density for all materials classes.

software (Granta 1999), which gives, among many performance polymers can operate up to 250 mC,
other properties, the maximum service temperatures. copper alloys and nickels are also suitable. For
This allows one to compare maximum operating temperature ranging between 400 mC and 575 mC, low-
temperatures for materials from various classes, the alloy ferritic steels, titanium alloys, and inconels can
abscissa being density for the sake of clarity in the operate. Between 575 mC and 650 mC, ferritic stainless
Fig. 2. The simplest comparison is that polymers have steels, austenitic stainless, steels and nimonics are
relatively low maximum service temperatures (less good candidates, whereas between 650 mC and 1000 mC,
than 100 mC for most of them), ceramics have operating austenitic stainless steels, nickel-based superalloys,
temperatures up to 1800 mC, and metals have operating and cobalt-based superalloys are suitable. Above
temperatures between room temperature and about 1000 mC, refractory metals such as molybdenum and
1500 mC. Figures 3, 4, and 5 show a similar map tantalum, and ceramics are the appropriate choice.
for metals and alloys, ceramics, and polymers, re- These very rough statements need, however, to be
spectively. On the horizontal axis is the reference supplemented with detailed creep data in order to go
temperature: melting temperature for metals and further in the comparison procedure.
ceramics, glass temperature for polymers. Whereas
the maximum temperature of metals and ceramics cor-
relates strongly with their melting temperature, the 5. Format of the Creep Data Suitable for
correlation for polymers with the glass transition Comparison Purposes
temperature is far less obvious, and this quantity is, One of the main difficulties in comparing materials for
thus, not a good guideline for estimation. creep resistance is to have the data available in a
similar format. Very often, this is not the case, and
depending on the materials, the tradition of data
4. Classes of Materials for Different Temperatures
recording is different. For polymers, one often finds
These maximum service temperatures, though a very the time-dependant modulus or the compliance as
rough evaluation of creep behavior, still provide one information on creep behavior. For metals and alloys,
with a simple classification of materials in terms of the one finds often the stress necessary to obtain failure in
operating temperatures aimed at. When the tempera- less than a given time. These data are perfectly suitable
ture range is 0–150 mC, polymers (most of them up to for some applications, but they are not adapted to the
60 mC, some of them up to 150 mC), magnesium, and general purpose of design, and they are difficult to
aluminum alloys can operate relatively safely. For extrapolate to other conditions. Therefore, it is pro-
temperatures between 150 mC and 400 mC, some high- posed that a uniform phenomenological description of

3
Creep Behavior of Materials: A Comparison

Figure 3
Maximum service temperature as a function of the melting temperature for metals and alloys.

creep data such as the one used for alloys could be provided by a power law as a function of time, with an
more adapted for the purpose. The standard way of exponent less than unity. Provided that one has the
describing steady-state creep is the so called Norton creep data in this format, one can compare the
law, valid for a range of stresses and temperatures, and materials for a given set of requirements, which now
defining the stress exponent (n) and the activation comprise a time-dependant deformation.
energy (Q):
E G E G
σ n Q 6. Performance Indices for Resistance Against
εc l A exp k (1) Steady-State Creep
σ RT
F !H F H
In order to simplify the approach, we will first assume
The time to rupture, tR, usually follows a similar law: that the material is in a steady-state creep condition
E G E G that follows Norton’s law. Consider as an example the
σ −n Q
tR l B exp (2) simple case of a clamped beam, with a length L and a
σ H RT H square section a#, which has to sustain a load F at one
F ! F
end, so that the bending in a given time t* does not
where σ is a reference stress. Creep damage tolerance exceed a prescribed value δ*. This design requirement
!
is approximately given by the parameter λ: is to be fulfilled with the aim of minimizing the mass.
Under these assumptions, the creep bending of the
εR
λl (3) beam is given by
εc stR
E G E G
2 kQ 4F 1 2nj1
where εR is the strain to fracture, and εs the minimum δc l A exp L#
nj2 RT σ a$ "
( n+ )/n 2n
strain rate. F H F ! H

For a limited range of temperatures and stresses, the (4)


creep ductility and the creep damage tolerance are
reasonably constant, in a range between 0.05 and 3 for This quantity has to remain lower than δ *\t *, which
ductility and between 1 and 3 for creep damage prescribes a minimum value for a. This in turns
tolerance. imposes that the mass be larger than
The mathematical formulation of primary creep is E G
#/$
4F L
less transparent. For very low stresses, it is generally m l ρL (5)
logarithmic, for higher stresses, a good description is F
σ* H

4
Creep Behavior of Materials: A Comparison

Figure 4
Maximum service temperature as a function of the melting temperature for ceramics.

The ‘‘performance index’’ for this problem, i.e., the A detailed list of performance indices for creep is to be
combination of properties to be maximized in order to found in the CES software (Granta 1999). It has to be
get an efficient design, is the quantity stressed that this method allows a comparison of
materials for creep resistance in a given set of
σ * #/$ requirements, provided the data of Norton’s law are
Il (6) available. By contrast with selection methods for room
ρ
temperature design, such data are not yet available in
design-oriented databases, so that the procedure has
Formally, it is exactly the same expression as if the
to be done manually.
design was made for strength at room temperature,
but replacing the yield strength by the quantity σ *,
which measures the creep strength of the material.
This creep strength is defined by 7. Case of a Nonsteady-state Creep
Very often, especially when one has to design with
A
nj2 δ* 1
C
"/n polymers, one is facing creep problems that are not
σ* l σ (7)
! B 2 t* L A exp (kQ\ RT ) D
steady state, and therefore not described by a Norton
type of law. Still, one can use a similar method,
Whereas the yield stress is a property of the materials although it cannot be dealt with analytically. Consider
alone, the creep strength involves not only materials a component under a tensile loading F, the aim still
properties (such as the parameters of Norton’s law) being mass minimization, the section of the component
but also design requirements such as the time, the being the free variable. The stress inside the component
maximum authorized deflection, and the length of the is given by:
beam. This method is very general and the perform-
ance indices for creep appear to be identical to the ones FLρ
given in Fig. 1, with an appropriate definition of creep σl (9)
m
strength. For instance, in case of a tensile loading, the
creep strength is
Let us imagine that we have, in a database, the strain
A C
"/n vs. time curves for a range of materials in a range of
δ*\t*
σ* l σ (8) temperatures and applied stresses σ , σ , σ , σ , etc.
! L A exp (kQ\ RT ) " the
Then for each material one can estimate # maximum
$ %
B D

5
Creep Behavior of Materials: A Comparison

Figure 5
Maximum service temperature as a function of the glass transition temperature for polymers and composites.

stress that it can sustain without reaching in the been selected, requires more advanced techniques such
prescribed time the maximum deformation allowed. as the ones listed in Ashby and Abel (1995).
This will in turn provide us with a maximum mass of
the component allowing for a comparison between the
different materials.
Bibliography
Ashby M F 1999 Materials Selection in Mechanical Design.
8. Conclusions Butterworth Heinemann, Oxford
Ashby M, Abel C 1995 Materials design to resist creep. Phil.
In the early stages of design a comparison between Trans. R. Soc. A 351, 451
different materials has to be very open and has to ASM Engineered Materials Handbook 1991 Ceramics. ASM
consider very different materials. Creep design in this International, Metals Park, OH, Vol. 4
respect is especially challenging. The first comparison Granta 1999 CES software. Granta Design
to be made is the operating temperature of the device Penny R K, Marriot D L 1995 Design for Creep. Chapman and
Hall, London
with the maximum working temperature of the ma- Saechtling International Plastics Handbook 1983 Macmillan,
terials. This first screening allows one to shortlist a London
class of materials. Then, provided the data are avail- Smithells Metals Handbook 1992 Butterworth Heinemann,
able in the same format, the performance index Oxford
method generalized with the concept of creep strength Waterman N, Ashby M (eds.) 1996 Chapman and Hall Materials
can allow for an objective comparison between ma- Selector. Chapman and Hall, London, Vol. 3
terials undergoing possible steady-state creep. Further
advance in design, once the appropriate material has Y. J. M. Brechet

6
Creep Behavior of Materials: A Comparison

Copyright ' 2001 Elsevier Science Ltd.


All rights reserved. No part of this publication may be reproduced, stored in any retrieval system or transmitted
in any form or by any means : electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or
otherwise, without permission in writing from the publishers.
Encyclopedia of Materials : Science and Technology
ISBN: 0-08-0431526
pp. 1759–1765

You might also like