You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/291336278

A NEW CREEP DAMAGE CRITERION AND ITS IMPLICATIONS TO


ENGINEERING CREEP DESIGN

Conference Paper · February 2006

CITATION READS

1 221

3 authors, including:

C. Phaniraj Baldev Raj


Indira Gandhi Centre for Atomic Research (DAE), Kalpakkam-603102, India Indira Gandhi Centre for Atomic Research
46 PUBLICATIONS   482 CITATIONS    918 PUBLICATIONS   16,565 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermography View project

Metal Matrix Nanocomposites View project

All content following this page was uploaded by C. Phaniraj on 28 March 2017.

The user has requested enhancement of the downloaded file.


Invited paper "Proceedings of Inter. Conf. on Pressure Vessels and Piping", OPE 2006 – CHENNAI, 07–09 Feb 2006,
Chennai, India.

A NEW CREEP DAMAGE CRITERION AND ITS IMPLICATIONS TO


ENGINEERING CREEP DESIGN

C. Phaniraj*, B. K. Choudhary and Baldev Raj


Metallurgy and Materials Group, Indira Gandhi Centre for Atomic Research,
Kalpakkam – 603 102, Tamil Nadu, India
(*e-mail: phani@igcar.gov.in)

ABSTRACT
In this paper, we introduce a new concept called time to reach Monkman–Grant ductility (i.e.,
MGD is the product of minimum creep rate and rupture life) tMGD, as the time at which the
useful secondary creep ductility is exhausted and 'true tertiary creep damage' sets in. For any
damage mechanism, it is put forward that tMGD is the time at which creep damage attains a
critical level. This is shown for a typical case of cavitation damage mechanism, where time to
attain critical cavity size matches well with tMGD. We propose a new critical creep damage
criterion in terms of a unique relationship between tMGD and rupture life that depends only on
the damage tolerance factor. The validity of the proposed creep damage criterion is shown for
9Cr–1Mo steel and AISI 304 stainless steel, the materials that are of interest to pressure
vessels and piping. Further, the universal nature of this criterion is demonstrated for a wide
range of materials from pure metals to ceramics and complex engineering alloys. This paper
addresses that it is appropriate to redefine the engineering creep design criterion in terms of
tMGD rather than rupture life. Further, we show that the damage criterion bears useful
implication to cumulative life fraction rule.
Keywords: Creep damage; Monkman–Grant ductility; Damage tolerance factor; Creep design

1. INTRODUCTION
The materials that are used for elevated temperature applications like boilers and pressure
vessels demand tolerance to creep damage. Creep damage is the progressive reduction in the
material’s ability to resist stress, and the initiation and growth of damage manifests as an
increase in deformation rate during tertiary creep stage eventually leading to failure. It is
desirable that the material spends most of its useful time in the secondary creep stage. The
inter-relationship between minimum (or steady state) creep rate  m and rupture life tr are
described by the well known empirical Monkman–Grant [1] (  m .tr = constant = CMG) and
modified Monkman–Grant [2] (  m .tr/r = constant = CMMG) relationships; where r is the
strain to rupture or failure.
The engineering creep oriented design criterion [3,4] in arriving at the allowable stresses
for nuclear applications is essentially based on the concept of deducing the lowest of the
three; i.e., 100% of the average stress to produce 1% strain in a given time (say 105 h), 67% of
the minimum stress to produce rupture in 105 h and 80% of the minimum stress to cause
initiation of tertiary creep in 105 h. The 1% strain criterion in 105 h, in essence, is equivalent
to the minimum creep rate of the 107 h–1 and this criterion safeguards the material against
distortion. Whereas, the second criterion is to guard against rupture or failure and the third is
to make sure that the material does not enter in to the accelerating tertiary creep stage which
eventually leads to the failure of material. Since, minimum creep rate is related to rupture life
(Monkman–Grant relation) and the time to onset of tertiary creep is also related to rupture life
[5], it essentially narrows down to a single criterion, i.e. rupture criterion. It may be noted that

1
in design, creep ductility is not considered while arriving at the allowable stress and this is
one of the important issues addressed in this paper.

2. BACKGROUND
The Monkman–Grant and modified Monkman–Grant relationships that are valid for most
of the materials suggest that the creep deformation and damage are not separate processes, but
are interrelated. They have also contributed significantly to the development of creep fracture
models [6]. In our recent work [5,7-9], we have placed the empirical Monkman-Grant and
modified Monkman-Grant relations on a firm and physically sound platform of first order
kinetics for transient and tertiary creep. It is important to mention that these relationships do
not describe the evolution of creep damage and its coupling to the deformation rate. This
coupling is described by the well known two approaches. The first is the ‘continuum creep
damage mechanics’ (CDM) approach of Kachanov and Rabotnov [10,11], where damage is
treated as an internal state variable. The most important outcome of CDM approach is the
creep damage tolerance factor  [12-16] which is defined as the ratio of strain to failure to the
product of  m and tr, i.e.,  = r / (  m .tr). For engineering alloys, its value [12] range from 1 to
about 20. It is a significant parameter that assesses the susceptibility of a material to localised
cracking at strain concentrations and for engineering components, it is suggested [14] to be a
better measure of creep ductility as it is related to the ability of a material to redistribute the
stresses. In simple terms, it is a measure of the tolerance of the material against creep damage;
a large  is desirable as the material can tolerate strain concentrations without local cracking.
Thus  is a material performance characteristic. It may be noted that  is reciprocal of the
modified Monkman-Grant constant CMMG and like CMMG,  also is a constant. The CDM
approach has received the status of physical basis ever since the pioneering work by Ashby
and Dyson [12,17]. The second approach that describes the coupling between creep damage
and deformation is the ‘Materials Properties Council (MPC)–Omega’ approach developed by
Prager [18]. According to this approach, creep rate  increases with strain  from its initial
value  0 as  =  0 exp(p ), where p is reciprocal of Monkman-Grant constant CMG, i.e.,
p = 1/(  m .tr) when primary creep is negligible (i.e., the initial creep rate  0 for undamaged
material can be approximated to  m when primary creep is negligible). The total damage
coefficient p is the rate at which material’s ability to resist stress is degraded by strain and is
also a material performance characteristic, i.e., higher the p, lesser is the resistance to creep
damage.
From the discussion presented so far, it is clear that Monkman-Grant relation is at the heart
of the problem. In this paper, henceforth we refer to Monkman-Grant constant CMG (i.e., the
product of  m and tr) as ‘Monkman-Grant ductility’ (MGD). MGD is the total secondary
creep strain contribution of strain to rupture r and is also the minimum creep ductility that a
material would posses under any embrittling creep damage situation. Thus MGD is the useful
secondary creep ductility and can be regarded as the intrinsic ductility available in a creeping
material. So, we define time to reach MGD, tMGD as the time at which MGD is reached along
the creep curve and the true tertiary creep damage sets in, since the useful total secondary
creep ductility is exhausted at this time. This paper addresses at what time the creep damage
that grows along the creep curve reaches a critical level. We put forward that t MGD is the time
at which creep damage attains a critical level and show this for a typical case of cavitation
damage mechanism in iron. Further, a critical damage criterion is deduced, following
CDM and MPC–Omega approaches, in terms of a unique relationship between tMGD and tr that
depends only on a single parameter . We also show that this relationship between tMGD and tr

2
can be derived following the Theta projection concept [19,20]. The validation of the critical
damage criterion and its useful implications to tertiary creep damage and engineering creep
design are also discussed in this paper. The concept of tMGD as the time at which critical
damage is attained is shown for creep cavitation mechanism in the following section.

t = (-1) m.tr
.
Creep strain ()

r =  (m.tr)
.

m.tr = MGD
.
p

tot tMGD tr
Time (t)

Fig. 1. Schematic creep curve illustrating negligible primary creep strain p,
time to reach Monkman-Grant ductility tMGD, time to onset of tertiary creep
tot, damage tolerance factor  and limiting tertiary creep strain t.

3. TIME TO ATTAIN CRITICAL CREEP DAMAGE


The concept and the determination of tMGD is shown in Fig. 1, where tMGD can be obtained
as the time at which   p = MGD (i.e., p is the limiting primary creep strain in Fig. 1) and
for negligible p, tMGD is the time to reach MGD. Creep damage can occur by various
mechanisms and if measurement of damage is carried out along the creep curve, then the
damage would attain a critical level at tMGD, beyond which its accelerated growth leads to
failure. This need to be validated and the mechanistic data (including that at tMGD) are
necessary which is hard to find in the literature. In the following, we analyse the mechanistic
data reported by Davis and Williams [21] on -iron for the creep tests conducted at
temperatures ranging from 815 to 978 K and in the stress range 17.24– 68.95 MPa. They
observed that the tertiary creep consisted of two stages. In the first stage, tertiary creep strain-
time followed a t4/3 law, whereas the second stage obeyed an exponential law. They reported
that the end of first stage corresponds to the time at which creep cavities attain a critical size,
i.e., in other words they reach a thermodynamically stable size and their further growth caused
the acceleration in creep rate. Hence, they concluded that the end of first stage or the
beginning of second stage is the ‘true tertiary creep’ that is marked by the cavities reaching a
critical size. Accordingly, in this paper, this time corresponding to the end of first stage is
designated as the time to attain critical cavity size tCCS. Creep curves that include secondary
and tertiary creep regimes were generated using the creep data as well as the tertiary creep

3
strain-time relations given by Davis and Williams [21]. f was obtained as the sum of p,
MGD and t (cf. Fig. 1, t is the limiting tertiary creep strain), where t was calculated

Time to attain critical cavity size (tCCS ), h 3


10  - Iron

2
10

1
tCCS = tMGD
10
815 K
866 K
0
10 923 K
978 K

-1
10 -1 0 1 2 3
10 10 10 10 10
Time to reach MGD (tMGD), h

Fig. 2. Plot of time to attain critical cavity size tCCS vs. tMGD for creep
cavitation micromechanism in -Fe. Solid line is according to tCCS =
tMGD and symbols correspond to the experimental data.

following their relation t = A exp[(tr  tot)] and tot is the time to onset of tertiary creep; the
values of A, , tot and tr are taken from their paper [21]. The constant reported value [22] of p
was taken as 0.051. From the generated creep curves,  (i.e., discussed later) and tMGD were
determined; tMGD for each creep curve was obtained as the time at which   p = MGD. The
plot of tCCS vs. tMGD for various temperatures and stress levels is shown in Fig. 2. This plot
clearly shows that tCCS matches well with tMGD and validates our proposition that tMGD is the
time at which creep damage attains a critical level for the cavitation damage mechanism. We
like to extend such a validation for other damage mechanisms, but it has not been possible due
to lack of detailed mechanistic data as well as the creep curve information. Further work in
this direction is needed to establish firmly this proposition, but this involves extensive
interrupted creep testing and microscopic examination. In the following section, the critical
damage criterion, in terms of a unique relationship between tMGD and tr, is deduced based on
CDM, MPC–Omega approaches and the theta method.

4. CRITICAL CREEP DAMAGE CRITERION


In this section, the critical damage criterion for creeping materials in the form of a unique
relationship between tMGD and tr that depends only on single parameter  is deduced. This is
obtained based on CDM, MPC–Omega approaches and the  projection method. It may be
noted that at longer times to failure and/or for the components that operate in the ASME
design stress levels, primary creep strain p is normally insignificant [16,18] compared with r
and, for the present analysis, is neglected. The damage criterion obtained according to the
CDM approach is given in the following.

4
4.1 CDM Approach
From CDM approach, recently [23] we deduced a new relationship between tMGD and tr in
terms of , and the same is briefly described here for the sake of completeness. The material
deforms at an increasing creep rate  in the tertiary regime. According to CDM, the evolution
of deformation and damage is expressed as two coupled differential equations in terms of
internal state damage variable . For uniaxial stressing [13,15], the creep and damage rates
take the simplest form:  =  0 (/0)n [1/(1  )]n and  =  0 (/0)m [1/(1  )]m, where  0
and   0 are temperature dependent rate constants at a stress 0; n and m are constants. It is
assumed that  = 0 when the material is in its undamaged state and  = 1 at rupture.
Integration of these coupled equations at constant stress, for m + 1 > n gives relation between
strain fraction /r and time fraction t/tr as

/r = 1  (1  t/tr)1/, (1)

where the damage tolerance factor  = (m+1)/(m  n + 1). For negligible p, it follows from
Fig. 1 that t = (  1)  m .tr, since  = r /(  m .tr ) and t = r  (  m .tr). At t = tMGD, creep
strain  = MGD =  m .tr, and on substituting this in Eq. (1), and on further rearrangement gives
t/r = (1  tMGD/tr)1/. In this equation, substituting for t and , we obtain the 'critical damage
criterion' in terms of a universal relationship between tMGD and tr as


t MGD    1
 1   = constant = fCDM , (2)
tr   

where fCDM can be determined knowing the value of . It may be emphasised that for a given
material and for the stress–temperature domain,  is constant, and in turn fCDM is also a
constant. It is interesting to note that unlike tMGD and tr that vary with the test condition,
tMGD/tr is independent of stress and temperature. We call the physically based Eq. (2) as
‘critical damage criterion’ because creep damage attains critical value at tMGD when the
criterion tMGD = fCDM tr is met, whereas tMGD is just the time to attain critical damage. The
theoretical plot of tMGD/tr vs.  following Eq. (2) is shown in Fig. 3, where tMGD/tr decreases
with increasing  and saturates at 0.63. This is in order since Eq. (2) is of the functional form
y = 1  [1  (1/x)]x and in the limit x  , y = (1  1/e) = 0.63. In Fig. 3,  = 1 means that the
material fails without any tertiary creep, i.e., r =  m .tr.

4.2 MPC–Omega Approach


Along the similar lines of CDM approach, the critical damage criterion as a relationship
between tMGD and tr is also obtained following MPCOmega method developed by Prager
[18]. As described earlier in section 2, according to this approach [18], creep rate  in the
tertiary regime increases with  from its initial value of  0 as  =  0 exp(p ), where p
corresponds to the total damage coefficient (i.e., p = m + p + c) that gives rise for the
increase in creep rate due to cross-sectional (i.e., m) plus creep damage (i.e., p) plus other
microstructural effects (i.e., c). It is important to note that when primary creep is negligible
and in agreement with Prager, it is reasonable to assume  m instead of  0 and accordingly the
evolution of creep rate can be written as

5
1.0
-Fe [21] Al2O3 [30]
W [27] Si3N4 [31]
0.9 Mo [27] MoSi2-SiCp [32]
9Cr-1Mo - low stress [9,26]
9Cr-1Mo - high stress [9,26]
Cu [28]
0.8 -Fe [29]
tMGD/tr

AISI 304 SS [5,7]


Re and Re-Os-W Alloy [27]
0.7

0.6
CDM approach (Eq. 2)
MPC Omega approach (Eq. 6)
0.5
1 3 5 7 9 11 13 15 17 19
Damage tolerance factor ()

Fig. 3. Validity of critical damage criterion (Eq. 2) for a wide range of


materials. Solid line is according to Eq. (2) based on CDM approach,
whereas broken line is according to Eq. (6) based on MPC–Omega
method. Symbols correspond to EXP values obtained from double
logarithmic plot of tMGD vs. tr for various materials. Data taken from
Ref. 27: data for W is taken from Fig. 5.9, Mo from Fig. 8.13, Re and
ReOsW alloy from Fig. 6.5 and Fig. 6.17B, respectively.

 =  m exp(p ) . (3)

On integration of Eq. (3), strain-time function can be obtained as

1
m  P

1  e   P  t .  (4)

For any time t and at tr (i.e., at failure), Eq. (4) leads to that given by Prager (i.e., Eq. 12 in
Ref. [18] with  m instead of  0 ) as

1
m  P
 
e  P  e  P r  t r  t  . (5)

It follows from Eq. (4) that p is the reciprocal of MGD, i.e., p = 1/(  m .tr), since at failure
(i.e., at tr), the exponential term in Eq. (4) is negligible [18] for large values (i.e., 2 or 3) of the
product p.
Starting from Eq. (5), the critical damage criterion in terms of the relationship between
tMGD and tr can be deduced. Since p = 1/(  m .tr) and from the definition of , it follows that
r = /p. Further, referring to Fig. 1, at t = tMGD, creep strain  = MGD =  m .tr for negligible

6
p. Substituting for p and r as well as the condition  = MGD =  m .tr at t = tMGD in Eq. (5),
and on rearranging, we obtain the unique relationship between tMGD and tr as

t MGD
 1  e 1  e  = constant = fMPC, (6)
tr

where fMPC can be determined knowing the constant value of . It may be noted from Eq. (6)
that unlike tMGD and tr, tMGD/tr is a constant independent of stress and temperature. Further,
when  >>1, it follows from Eq. (6) that tMGD/tr = 1 – (1/e) = 0.63 and this is in accordance
with Eq. (2) for large values of . The theoretical plot of tMGD/tr vs.  following Eq. (6) is also
shown in Fig. 3 illustrating that fMPC decreases with increasing  and saturates at 0.63. Thus
MPC-Omega method further strengthens our relationship obtained according to CDM
approach (i.e., Eq. 2) and reinforces that tMGD/tr depends only on .

4.3 Theta Projection Method


In this section, the relationship between tMGD and tr following the  projection method is
obtained. In a recent paper, Wilshire and Lieu [24] have reported constant stress tensile creep
behaviour for the oxide-dispersion-strengthened (ODS) ferritic steel, Incoloy MA957 at 873 –
973 K and at stresses ranging from 250 to 420 MPa. They have presented an interesting
analysis of the shape of creep curves for MA957 in terms of  projection concept [19,20] that
describes the saturating primary and accelerating tertiary creep in terms of strain–time
relationship as

 = 1 [1 – exp(– 2t)] + 3 [exp(4t) – 1] , (7)

where 1 and 3 quantify primary and tertiary strains, while 2 and 4 are rate parameters that
govern the curvatures of the primary and tertiary components, respectively. They found that
creep curves over most of the creep life were accurately described by the  methodology. It
follows (cf. Fig. 12 in Ref. [24]) that 1  p which is the strain intercept at t = 0 obtained by
back-extrapolation of the minimum creep rate line. Wilshire and co-workers [24,25] have
observed important correlation between the  parameters and,  m and tr as

 m = C134, (8)

4tr = C2, (9)

where C1 and C2 are constants with C1  1 and C2  1.


For negligible p, Eq. (7) can be expressed as  – p = 3[exp(4t) – 1] since p = 1, and
substituting the condition  – p =  m tr = MGD at t = tMGD, Eq. (1) takes the form

 m tr = 3[exp(4tMGD) – 1]. (10)

From Eqs. (8) and (9), it follows that  m tr = C1C23. Substituting this in Eq. (10) and on
rearrangement, we get 4tMGD = ln [1 + (C1C2)]. Further, with Eq. (9), knowing 4 = C2/tr, the
relationship between tMGD and tr can be easily deduced as

7
tMGD/tr = ln[1 + (C1C2)] / C2 = constant = fPC, (11)

where the subscript ‘PC’ means  projection concept. The same conclusion is obtained from
Eq. (11) that though tMGD and tr vary with test condition, the ratio tMGD/tr is constant
independent of stress and temperature, since C1 and C2 are constants. Following  projection
method, we could not get the relationship between tMGD and tr directly in terms of  (i.e., as
obtained through CDM and MPC–Omega approaches). However, it may be noted that  is
related to the constants C1 and C2 [24,25].

5. EXPERIMENTAL VALIDATION OF CRITICAL DAMAGE CRITERION


The validity of critical damage criterion (i.e., Eqs. 2 and 6) is demonstrated for a wide
range of materials. First, its experimental validity is shown for the materials that are of
interest to pressure vessels. We use our constant load creep test data obtained on 9Cr1Mo
ferritic steel [8,9,26] in quenched and tempered (Q+T) condition at 793 and 873 K in the
stress range of 60275 MPa, and on AISI 304 stainless steel [5,7] in air and argon
environments at 973 K in the stress range of 100200 MPa. Both these steels showed
negligible p compared to strain to failure and two stress regimes with different stress
exponents for dependence of creep rate were observed.  was determined as the intercept at
 m = 1 from the double logarithmic plot of average creep rate r /tr vs.  m , since it follows
from the definition of  that r /tr =   m . Figure 4 shows constancy of  for 9Cr1Mo steel in
the respective stress regimes. Also plotted in Fig. 4, is the published data obtained by Davis
and Williams for iron [21]. For iron, the value of  = 1.6 was obtained, whereas
9Cr1Mo steel exhibited separate values of  = 10 and 5 for low and high stress regimes,
respectively. It may be noted that for 9Cr1Mo steel, low stress regime corresponds to the
data obtained at 873 K, 6090 MPa, whereas high stress regime is that at 793 K (150275
MPa) and 873 K (100175 MPa). Figure 5 shows that  = 2.1 for AISI 304 stainless steel in
both air and argon environments.

0
10 AISI 304 Stainless Steel
 - Fe
-1
Average creep rate (r /tr), h
-1

-1
9Cr-1Mo steel - high stress 10
Average creep rate (r/tr), h

-1 9Cr-1Mo steel - low stress


10  = 1.6
Temperature - 973 K
=5 -2 Air (120 - 200 MPa)
-2 10 Argon (100 - 200 MPa)
10  = 2.1

-3
10 -3
10
 = 10
-4
10
-4
10
-5
10 -4 -3 -2 -1
10
-6
10
-5
10
-4
10
-3 -2
10
-1
10
0
10 10 10 10 10
. . -1
Minimum creep rate (m), h
-1
Minimum creep rate (m), h

Fig. 4. Illustrates the constancy of  Fig. 5. Constancy of  for AISI 304


for 9Cr1Mo steel and iron. stainless steel.

8
The validity of the critical damage criterion obtained based on CDM approach (i.e., the
relationship between tMGD and tr in terms of  given by Eq. 2) is demonstrated in Figs. 69 for
iron, 9Cr1Mo steel and AISI 304 stainless steel. In these figures, symbols represent the
experimental data points and the solid line corresponds to the theoretical line according to Eq.
(2) with respective value of ; i.e., tMGD = CDM tr and CDM = 1  [(  1) /]. Figure 6
shows the validity of the damage criterion for -iron with CDM = 0.79 determined using  =
1.6. The validity of the damage criterion for low and high stress data of 9Cr-1Mo steel is
shown in Figs. 7 and 8, respectively; whereas Fig. 9 corresponds to the data obtained on AISI
304 stainless steel. For AISI 304 stainless steel CDM = 0.74 (Figure 9), whereas for 9Cr-1Mo
steel it is 0.65 (Fig. 7) and 0.67 (Fig. 8) for low and high stress regimes, respectively. It is
important to note that the analysis of creep data in terms of tMGD vs. tr plots also obeyed the
damage criterion deduced based on MPC–Omega method (i.e., Eq. 6) for -iron, 9Cr-1Mo
steel and AISI 304 stainless steel. From the plots of tMGD vs. tr, the observed values of
constant of proportionality EXP (i.e., tMGD = EXP tr) were also determined for the experimental
data and these are shown as symbols in Fig. 3 illustrating the validity of the creep damage
criterion obtained by both CDM and MPC–Omega approaches.

5
3 10
10  - Iron 9Cr - 1Mo Steel (Q+T)
Time to reach MGD (tMGD), h
Time to reach MGD (tMGD), h

815 K
2 Temperature - 873 K
10 866 K Low stress regime (60 - 90 MPa)
4
923 K 10
978 K
tMGD = fCDMtr
10
1 tMGD = fCDM tr
(fCDM = 0.79)
(fCDM = 0.65)
3
10
0
10

-1 2
10 -1 0 1 2 3
10 3 4 5
10 10 10 10 10 10 10 10
Rupture life (tr), h Rupture life (tr), h

Fig. 6. Variation of tMGD with tr Fig. 7. Plot of tMGD vs. tr showing the
demonstrating the validity of critical validity of critical damage criterion
damage criterion (Eq. 2) for -Fe with (Eq. 2) for 9Cr-Mo steel at low
CDM value calculated using  = 1.6. stresses with CDM value calculated
Symbols represent experimental data using  = 10. Symbols represent
obeying Eq. (2). experimental data obeying Eq. (2).

Other than above materials, the validity of the damage criterion is demonstrated for a wide
range of materials. The published tensile creep data [2732] for various materials such as pure
metals, ceramics and composite of intermetallic silicide were analysed for , tMGD and tr. 
was found to be constant for the given material and this for a typical case of published data on
iron is already shown in Fig. 4. For different materials, logarithmic plots of tMGD vs. tr were
drawn, and they obeyed tMGD  tr (i.e., tMGD = fEXP tr), and the observed fEXP values are shown

9
as symbols in Fig. 3 illustrating the validity of the damage criterion. Further, the plot of tMGD
vs. tr in Fig. 10 shows the validity of damage criterion (Eq. 2) for various materials with
respective fCDM values; the fCDM ranged from 0.65 for 9Cr1Mo steel to 0.91 for Al2O3.
Unlike that reflected in Fig. 3, the difference in fCDM values is not seen in Fig. 10 due to
logarithmic representation of the plot. An interesting observation to make is that the damage
criterion is valid for the range of rupture life data from approximately 0.04 to 104 h.

9Cr - 1Mo Steel (Q + T) AISI 304 Stainless Steel


Time to reach MGD (tMGD), h

4
10

Time to reach MGD (tMGD), h


3
10
High stress regime Temperature - 973 K
3
10 793 K, 150 - 275 MPa Air (120 - 200 MPa)
873 K, 100 - 175 MPa Argon (100 - 200 MPa)
2
tMGD = fCDMtr 10 tMGD = fCDMtr
2
10 (fCDM = 0.67) (fCDM = 0.74)

1 1
10 10

0
10
0
10 1 2 3 4
0 1 2 3 4
10 10 10 10 10 10 10 10 10
Rupture life (tr), h Rupture life (tr), h

Fig. 8. Validity of damage criterion Fig. 9. Validity of damage criterion


(Eq. 2) for 9Cr-Mo steel at high (Eq. 2) for AISI 304 stainless steel
stresses with CDM value calculated with CDM value calculated using  =
using  = 5. Symbols represent 2.1. Symbols represent experimental
experimental data obeying Eq. (2). data obeying Eq. (2).

5
10
fCDM=
 -Fe
t (h)

4 0.79
10 W 0.75
Time to reach MGD MGD

Mo 0.78
3 9Cr-1Mo - 0.65
10
low stress
2 9Cr-1Mo - 0.67
10 high stress
Cu 0.81
10
1  -Fe 0.80
AISI 304 SS 0.74
fCDM=
0
10 Re and 0.75
Re-Os-W Alloy
10
-1 Al2O3 0.91
Si3N4 0.77
-2 MoSi2-SiCp 0.72
10
-2 -1 0 1 2 3 4 5
10 10 10 10 10 10 10 10
Rupture life tr (h)

Fig. 10. Plot of tMGD vs. tr demonstrating the validity of damage


criterion (Eq. 2) with CDM values calculated using the respective
values of  for different materials. Symbols represent experimental
data obeying Eq. (2).

10
A comment on the validity of the critical damage criterion in terms of the relationship
between tMGD and tr is in order. Ashby and Dyson [12] have demonstrated that each damage
micromechanism, when acting alone, is reflected in a characteristic shape of creep curve and a
correspondingly characteristic value of . Therefore, Eqs. (2) and (6) are valid when the
damage mechanism does not change, as any change in mechanism results in change in the
value of  that in turn changes the value of fCDM (or fMPC). However, for situations when 
varies and for values of  > 4, it can be seen in Figure 3 that fCDM (or fMPC) almost remains
constant indicating the applicability of Eqs. (2) and (6). Further, while deducing Eqs. (2) and
(6) [or Eq.11], we only substituted the condition of  =  m .tr = MGD at t = tMGD and did not
invoke the constancy of MGD (i.e., CMG in the Monkman-Grant relation need not be
constant). This implies that the damage criterion given by Eq. 2 (or Eqs. 6 or 11) is valid
whether MGD is constant or varies with test condition; since it is only dependent on . In this
context, it is worthwhile to recall the creep data of Wilshire and Lieu [24] for Incoloy
MA957. Using their reported values of C1  1 and C2  1 for MA957 in Eq. (11), fPC can be
calculated and fPC = 0.7 independent of stress and temperature. They found the constancy of
 and   1.5. Following Eq. (2), fCDM can be determined knowing   1.5 and fCDM = 0.8
which is comparable to that obtained from  approach. In their study, MGD was not constant,
but found to vary with test condition (cf. Fig. 17 in Ref. [24]); MGD decreased with decrease
in applied stress particularly at higher test temperatures and MGD ranged from 0.01 to 0.18.
Thus, the damage criterion is valid even if MGD varies with the test condition, as it is
dependent only on .

6. IMPLICATIONS OF DAMAGE CRITERION TO TERTIARY CREEP AND


ENGINEERING CREEP DESIGN
The critical damage criterion has important implications to tertiary creep damage and
engineering creep design. The proposal of redefining the creep design criterion in terms of
tMGD as well the implication of damage criterion to life fraction rule is discussed in this
section.

6.1 Implication to Tertiary Creep Damage


Conceptually tMGD divides the creep curve into two parts (Fig. 1) and this can be
considered as something similar to the tensile stress-strain curve, where the plastic instability
divides the stress-strain curve into uniform and non-uniform deformation regimes. However,
it should not be mistaken that necking during creep occurs at tMGD. The essential message of
the Monkman-Grant relation is that creep deformation and damage are interrelated and creep
fracture is strain controlled. For cavitation damage mechanism in iron, we have shown that
tMGD is equal to the time at which cavities attain a stable critical cavity size. The stage from
tMGD to failure corresponds to the linkage of cavities resulting in micro-cracks thus causing
accelerated creep damage and failure. In other words, it may be understood as the damage is
homogeneous till tMGD and it is beyond this time till failure is the region of localised creep
damage. For creep curves obtained at different stresses and temperatures (i.e., with constancy
of  in a stress–temperature domain), the points on the creep curves corresponding to tMGD
can be joined to obtain the contour that satisfies tMGD = fCDM tr. Then, the region below such a
contour can be considered as the safe region, whereas above this (i.e., region from t MGD to tr)
as the unsafe region governed by the accelerated growth of creep damage. Further, it may
more appropriate to redefine the design criterion for creep of elevated temperature

11
components in terms of tMGD rather than tr. The implications of damage criterion to creep
oriented engineering design are given in the following section.

6.2 Implication to Design Criterion for Creep


Another important implication of damage criterion is for engineering creep design
criterion. The criterion for arriving at the allowable stress [3,4] has already been described in
section 1. Stress to cause rupture in 105 h with a factor of safety of 67% is used as one of the
criterion to arrive at the design allowable stress intensity limits. Based on the experience, this
factor of safety is employed as a rule of thumb by the material engineers. It has been observed
in Fig. 3 that the ratio of tMGD/tr decreases with increase in  and saturates to a value ~ 2/3.
We suggest that this would provide a physical basis for the factor of safety of 67% that is
practiced as a rule of thumb. Further, as critical damage sets in and the minimum ductility is
assured up to tMGD, we suggest that the stress to cause tMGD in 105 h can be considered as a
new design criterion for components that operate in the creep range; however with a suitable
factor of safety that is less conservative than the rupture criterion. The advantage of redefining
the creep design criterion rather in terms of tMGD than tr is that the remaining life, at the first
instance, can be estimated as tr tMGD.

6.3 Implication to Life Fraction Damage Rule: Modified Rule


The damage criterion also bears its implication to life fraction damage rule [4,33,34]. For
non-steady stress and temperature conditions, conventional creep damage rules that have been
proposed in the literature are life fraction rule by Robinson [4,33,34], strain fraction rule by
Lieberman [35], mixed life and strain fraction rule by Voorhees and Freeman [36], and the
modified mixed criterion by Abo El Ata and Finnie [37] for multiaxial stress conditions.
Among these, Robinson’s life fraction rule is most widely used in engineering creep design
and life prediction of high temperature components. According to this rule, failure is predicted
to occur when

n
ti
t
i 1
 1, (12)
ri

where ti is the time spent at any given stress and temperature, tri is the rupture life under
those conditions and n is the number of stress or temperature changes. For n = 2, i.e., varying
load and/or temperature creep rupture tests comprised of only two stages, life fraction rule can
be written as

t1 t 2
  1, (13)
t r1 t r 2

where subscripts 1 and 2 correspond to first and second stages, respectively.


It has been observed [4,34,37] that the life fraction rule is valid for temperature changes,
but not for stress changes. Further, for conditions when ti/tri is either equal to or exceeds
unity, the life fraction rule can be conveniently used as a conservative basis for life prediction.
But, when ti/tri < 1 (i.e., life fraction rule fails), it predicts non conservative values of
remnant life. For such situations, based on the concept tMGD, a modified life fraction rule that
is conservative, is proposed as

12
n
t i
t i 1
 1, (14)
MGDi

where ti is the time spent at a particular stress and temperature, tMGDi is the value of tMGD
corresponding to that stress and temperature. For n = 2, the modified rule can be written as

t1 t2
  1, (15)
t MGD1 t MGD2

where the subscripts 1 and 2 correspond to first and second stages, respectively. This
proposed modified rule implicitly incorporates the strain-related term, as tMGD is the time at
which the useful Monkman-Grant strain (i.e., MGD) is exhausted. This concept of
understanding the creep damage in terms of strain is in accordance with our earlier work on
tertiary creep [5] and the recent study of creep damage evolution by Sasikala et al. [38]. It can
be seen that the above modified life fraction rule is conservative. It can be easily shown that
ti/tMGDi = 1 leads to ti/tri < 1 as tMGD = fCDM tr (i.e., ti/tMGDi = (1/fCDM) ti/tri = 1 and
ti/tri = fCDM < 1), since it is reasonable to assume that the fCDM remains constant (i.e.,  is
constant) in a given stress-temperature domain.
Studies pertaining to life fraction rule provide the experimental values of t and tr. But, to
verify the conservative nature of the proposed modified rule, tMGD values are also required and
such data are hard to find in the literature. In this context, the experimental data obtained by
Tsuji et al. [39,40] on a nickel base Hastelloy XR is re-examined to verify the modified life
fraction rule. Tsuji et al. [39,40] conducted constant load as well as varying load and/or

-2
10
Hastelloy XR Hastelloy XR
-1
Average creep rate (r /tr), h

1.0 1223 K
Temperature - 1223 K
-3
10 Stress range - 14.7-22.8 MPa
t2/tr2 or t2/tMGD2

-4  = 2.9
10
0.5

-5
10
t2/tr2 vs. t1/tr1
t2/tMGD2 vs. t1/tMGD1
-6 -5 -4 -3 0.0
10 10 10 10 0.0 0.5 1.0
. -1
Minimum creep rate (m), h t1/tr1 or t1/tMGD1

Fig. 11. Constancy of  for the creep data Fig. 12. Illustrates conservative nature of
obtained by Tsuji et al. [40] for Hastelloy modified rule (Eq. 15) and the the non-
XR. applicability of life fraction rule (Eq. 13).

13
temperature creep rupture tests comprised of two stages in the temperature range 11231273
K. They reported that life fraction rule showed better applicability compared to other rules.
However, it has been pointed out by Schafer [41] that for the data obtained under variable
load conditions at 1223 K, t2(calculated)/t2(measured) increases with increasing first life
fraction (i.e., L1 = t1/tr1) and attains as high as 10 or more when L1 > 0.7 thus demonstrating
a significant deviation from life fraction rule. This failure of life fraction rule has been agreed
by Tsuji [42] in reply to Schafer’s comment.
The constant load and varying load creep rupture data obtained by Tsuji et al. [39,40] at
1223 K and at stresses ranging from 14.7 to 22.8 MPa is analysed. The values of tr and r at
1223 K at different stresses were calculated according to the best fit equations given by them
(cf. Eqs. 3 and 7 in Refs. 39 or 40); tr = 1.550  108  3.891 and r = 26.519 log   12.766.
Further, from the plot of  m vs.  at 1223 K (cf. Fig. 3 in Ref. 40), we obtained the best-fit
equation as  m = 6.588  1013  7. 066 and according to this equation calculated  m at
different stresses. Figure 11 shows the logarithmic plot between average creep rate (i.e., r/tr)
and  m illustrating the constancy of  = 2.9 for Hastelloy XR at 1223 K. Creep curves are
necessary for obtaining tMGD values at different stresses at 1223 K, but are not available for
the data of Tsuji et al. Therefore, following Eq. (2), tMGD was calculated as tMGD = fCDM tr,
where fCDM = 1 – [( – 1)/ ] = 0.71 for the observed value of  = 2.9. For varying stress
creep rupture tests at 1223 K, using the experimental values of t1, t2 (cf. test no. 811 of
Table 4(a) in Ref. 39) and the tMGD values, time fractions t1/tMGD1 and t2/tMGD2 were
calculated. For the data at 1223 K, both life fraction and modified life fraction rules (i.e., Eqs.
13 and 15) are presented in Fig. 12 which shows that life fraction rule is not valid (i.e., ti/tri
< 1), and the modified rule is conservative; in this figure, solid symbols represent modified
life fraction rule, whereas the open symbols correspond to the life fraction rule. Thus for
situations when ti/tri < 1 as observed for varying stress tests on Hastelloy XR at 1223 K, the
modified life fraction rule can be conveniently used as a conservative basis for life prediction.

7. CONCLUSIONS
The concept of time to reach Monkman-Grant ductility (MGD) i.e., tMGD is introduced as
the time at which useful secondary creep ductility is exhausted and the true tertiary creep
damage sets in. It has been put forward that for any damage micromechanism, creep damage
evolves along the creep curve and attains a critical level at tMGD. This proposition is validated
for cavitation damage mechanism in iron and is shown that the time to attain critical cavity
size matches well with tMGD. A critical damage criterion is proposed in terms of a unique
relationship between tMGD and tr that depends only on damage tolerance factor . The damage
criterion is deduced based on CDM approach as tMGD / tr = 1  [(  1)/] and is shown to be
identical to that obtained according to MPC–Omega method as tMGD/tr = 1 – e–1 + e–. This
damage criterion is also derived following the  projection concept. The validity of the
damage criterion is shown for 9Cr1Mo steel and AISI 304 stainless steel, the materials of
interest to pressure vessels. Also, its universal applicability is demonstrated for a wide range
of materials from pure metals to ceramics and complex engineering alloys. Further, its useful
implications to tertiary creep damage and engineering creep design are discussed in the paper.
Based on the concept of tMGD, a modified life fraction rule is suggested that is shown to be
conservative. It is advocated that it is more appropriate to redefine the creep design criterion
in terms of tMGD rather than rupture life.

14
REFERENCES
1. Monkman F C, and Grant N J, Proc. ASTM (1956), Vol 56, p. 593.
2. Dobes F, and Milicka K, Metal Sci 10 (1976) 382.
3. ASME Boiler and Pressure Vessel Code, Section VIII, Division 1 and Section III,
Division 1, Subsection NH, Class I Components in Elevated Temperatures Service (2001).
4. Woodford D A, in Materials Selection and Design, ASM Handbook, ASM International
(1977), Vol 20, p. 573.
5. Phaniraj C, Nandagopal M, Mannan S L, Rodriguez P, and Kashyap B P, Acta Mater 44
(1996) 4059.
6. Cocks A C F, and Ashby M F, Prog Mater Sci 27 (1982) 189.
7. Phaniraj C, Nandagopal M, Mannan S L, and Rodriguez P, Acta Metall Mater 39 (1991)
1651.
8. Choudhary B K, Phaniraj C, Bhanu Sankara Rao K, and Mannan S L, Key Eng
Materials 171–174 (2000) 437.
9. Choudhary B K, Phaniraj C, Bhanu Sankara Rao K, and Mannan S L, ISIJ Inter
Supplement 41 (2001) S73.
10. Kachanov L M, Izv Akad Nauk, USSR, Otd Tekd Nauk (1958) Vol 8, p 26.
11. Robotnov Y N, Creep Problems in Structural Members, North Holland Publ Co,
Amsterdam (1969).
12. Ashby M F, and Dyson B F, in Advances in Fracture Research, (eds) Valluri S R, Taplin
D M R, Rama Rao P, Knott J F, and Dubey R, Pergamon Press, Oxford (1984), Vol 1, p 3.
13. Dyson B F, and Gibbons T B, Acta Metall 35 (1987) 2355.
14. Leckie F A, and Hayhurst D R, Acta Metall 25 (1977) 1059.
15. Leckie F A, and Hayhurst D R, Proc Royal Soc A 340 (1974) 323.
16. Cane B J, Int J Press Ves Piping 10 (1982) 11.
17. Dyson B F, J Press Vessel Tech 122 (2000) 281.
18. Prager M, J Press Vessel Tech 117 (1995) 95.
19. Evans R W, Parker J D, and Wilshire B, in Recent Advances in Creep and Fracture of
Engineering Materials and Structures, (eds) Wilshire B, and Owen D R J, Pineridge
Press, Swansea (1982) p 135.
20. Evans R W, and Wilshire B, Introduction to Creep, The Institute of Materials, London
(1993).
21. Davis P W, and Williams K R, Acta Metall 17 (1969) 897.
22. Webster G A, Cox A P D, and Dorn J E, Metal Sci J 3 (1969) 221.
23. Phaniraj C, Choudhary B K, Bhanu Sankara Rao K, and Baldev Raj, Scripta Mater 48
(2003) 1313.
24. Wilshire B, and Lieu T D, Mater Sci Eng A 386 (2004) 81.
25. Burt H, and Wilshire B, Metall Mater Trans 35A (2004) 1691.
26. Choudhary B K, Saroja S, Bhanu Sankara Rao K, and Mannan S L, Metall Mater Trans
30A (1999) 2825.
27. Conway J B, and Flagella P N, Creep-Rupture Data for the Refractory Metals to High
Temperatures, Gordon and Breach Sci Publ, New York (1971).
28. Parker J D, and Wilshire B, Mater Sci Eng 43 (1980) 271.
29. Hough R R, and Rolls R, Metal Sci J 5 (1971) 206.
30. Robertson A G, Wilkinson D S, and Caceres C H, J Am Ceram Soc 74 (1991) 915.
31. Tanaka T, Nakayama H, Okabe N, Yamamoto S, and Fukui S , in Fracture Mechanics of
Ceramics–Fracture Fundamentals, HighTemperature Deformation, Damage, and
Design, (eds) Bradt R C, Hasselman D P H, Munz D, Sakai M, and Schevchenko V Ya,
Plenum Press, New York (1992), Vol 10, p 473.

15
32. French J D, Weiderhorn S M, and Petrovic J J, in High Temperature Silicides and
Refractory Alloys, Mater Res Soc Symp, (eds) Briant C L, Petrovic J J, Bewlay B P,
Vasudevan A K, and Lipsitt H A, Materials Research Society, Pittsburgh (1994), Vol 322,
p 203.
33. Robinson E L, Trans ASME 60 (1938) 253.
34. Viswanathan R, Damage Mechanisms and Life Assessment of High Temperature
Components, ASM International, Metals Park, OH (1989).
35. Lieberman Y, Mettaloved: Term Obrabotka Metal 4 (1962) 6.
36. Voorhees H R, and Freeman J W, Wright Air Development Center, Technical Report
5758, Part II, ASTIA Document No. 207 (1959) p 850.
37. Abo El Ata M M, and Finnie I, Trans ASME, J Basic Eng 94 (1972) 533.
38. Sasikala G, Ray S K, and Mannan S L, Acta Mater 52 (2004) 5677.
39. Tsuji H, Tanabe T, Nakasone Y, and Nakajima H, J Nucl Materials 199 (1992) 43.
40. Tsuji H, Tanabe T, Nakasone Y, and Nakajima H, J Nucl Sci Tech 30 (1993) 768.
41. Schafer L, J Nucl Materials 203 (1993) 186.
42. Tsuji H, J Nucl Materials 203 (1993) 187.

16

View publication stats

You might also like