International Journal of Pressure Vessels and Piping 78 (2001) 963±976

www.elsevier.com/locate/ijpvp

Development and use of the R5 procedures for the assessment of defects in

high temperature plant
D.W. Dean*, R.A. Ainsworth, S.E. Booth
Structural Integrity Branch, British Energy Generation Ltd, Barnett Way, Barnwood, Gloucester GL4 3RS, UK

Abstract
The R5 procedures have been developed within the UK power generation industry to assess the integrity of nuclear and conventional plant
operating at high temperatures. Within R5, there are speci®c procedures for assessing components containing defects. These are largely based
on approximate reference stress techniques and are continuing to be developed. This paper describes the procedures for assessing the
incubation and growth of defects at elevated temperatures and includes examples of both experimental and ®nite-element validation of these
approaches. The use of these basic procedures is then illustrated by application to a typical high temperature plant component.
Alternative methods for assessing incubation and the early stages of creep crack growth are currently being developed within the R5
procedures and this paper ®nally describes one of these, the time dependent failure assessment diagram (TDFAD) approach. q 2002 Elsevier
Science Ltd. All rights reserved.
Keywords: Creep; Creep±fatigue; Crack growth; Assessment

1. Introduction procedure [6] except that a `creep toughness' is used instead

of the conventional fracture toughness and time dependent
High temperature defect assessment procedures were stress and strain parameters are required. The development
developed some time ago [1] and are now well established of both stress and strain based TDFAD approaches is
[2]. The R5 procedures [3] produced within the UK power discussed together with issues relating to the determination
generation industry are widely used in plant assessment and of `creep toughness' from creep crack growth tests.
have been incorporated into a British Standards Document
PD6539 [4] with an ASTM standard [5] being available for
generating the required creep crack growth data. 2. Basic R5 defect assessment procedure
Con®dence in the basic R5 procedures has been
established to the extent that they are routinely used within The background to the defect assessment procedures in
British Energy for assessments of operating high tempera- R5 and PD6539 has been set out in Ref. [2]. Here, the basic
ture plant. Following a brief description of the basic defect procedures for steady loading of homogeneous components
assessment procedures in Section 2, examples of both are brie¯y presented. Then the treatment of welds and the
experimental and ®nite-element validation of these inclusion of fatigue crack growth are discussed. The
approaches are presented in Section 3. Section 4 then illus- procedures use approximate techniques based on a reference
trates the use of the basic R5 procedures by application to an stress related to the applied load P, by
Advanced Gas Cooled Reactor component operating at high
temperature; this includes a comparison of R5 predictions of s ref ˆ Ps Y =PL s Y ; a† 1†
creep±fatigue crack growth with measurements. Finally,
Section 5 describes recent developments of the TDFAD where PL is the plastic collapse load for a yield stress s Y and
approach for assessing incubation and the early stages of crack size a.
creep crack growth. This method is similar to that adopted in Having evaluated the reference stress, four times are
the established R6 low temperature defect assessment estimated. First the time tCD for creep damage to propagate
through a structure and lead to failure is taken as

* Corresponding author. Tel.: 144-1452-652-968; fax: 144-1452-653- tCD . tr ‰s ref a0 †Š 2†

025.
E-mail address: [email protected] (D.W. Dean). where tr(s ) is the rupture time at stress s , from conventional
0308-0161/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0308-016 1(01)00111-9
964 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

stress/time-to rupture data and the reference stress is calcu- by Ainsworth and Budden [10] as:
lated for the initial crack size a0.
Time is required for stress redistribution due to creep C t† 1 1 t=tred †n11
ˆ 8†
from the initial elastic state. Redistribution is complete Cp 1 1 t=tred †n11 2 1
and widespread creep conditions are established at a for a material obeying a Norton secondary creep law of the
redistribution time tred [7], which may be expressed form
conveniently in terms of the reference stress [8] as
e_ c ˆ Ds n 9†
e c ‰s ref a0 †; tred Š ˆ s ref a0 †=E 3†
For more generalised application Eq. (8) may be
where ec ‰s ref ; tŠ is the accumulated creep strain at the refer- expressed as:
ence stress for time t, from uniaxial creep data.
Next, the initiation time ti, during which the initial crack C t† 1 1 ecref =eeref †1= 12q†
ˆ 10†
blunts without any signi®cant crack extension is estimated. Cp 1 1 ecref =eeref †1= 12q† 2 1
The testing standard [5] does not include methods for where ecref is the accumulated creep strain at the reference
obtaining initiation data, but recent recommendations [9] stress after time t, eeref is the elastic strain at the reference
propose that initiation is de®ned for engineering purposes stress and q < n= n 1 1†† is the exponent in the creep crack
as corresponding to 0.2 mm crack extension. The method growth law (Eq. (11) below).
for representing initiation data then depends on observed The ®nal time required is the time tg, for the crack to
specimen response. For steady state creep conditions with propagate by an amount Da. For steady state creep this is
an essentially constant displacement rate, the initiation time obtained from creep crack growth data in the form
in test specimens is correlated with experimental estimates
of the crack tip parameter C p through a_ ˆ AC pq 11†

ti C pq ˆ B 4† where A and q are constants, using Eq. (7) to estimate C p for

crack sizes between a0 and a0 1 Da. To allow for transient
where B and q are constants. More generally, initiation creep and the increased amplitude of the crack tip ®elds at
times can be related to measurements of a critical crack short times estimated using Eq. (10), it is assumed that for
opening displacement d i, which can then be used to times less than the redistribution time t , tred † Eq. (11)
calculate a critical reference strain as may be generalised to

ec ‰s ref a0 †; ti Š ˆ ‰di =R 0 a0 †Šn= n11† 2 s ref a0 †=E 5† a_ ˆ A‰C t†Šq 12†

where the characteristic length R 0 is de®ned by: For situations where ti 1 tg . tred ; the effects of the redis-
tribution period can be allowed for by using the crack
R 0 ˆ K=s ref †2 6† growth rates of Eq. (11) multiplied by a factor of 2 for t ,
tred ; i.e.
As both K and s ref are directly proportional to the loading P,
the value of R 0 is independent of the magnitude of P. a_ ˆ 2A C p †q for ti # t , tred
13†
However, R 0 does vary with crack size and, when creep a_ ˆ A C p †q for t $ tred
crack growth is being considered, both K and s ref should
be calculated for the defect size equal to the size of the If the total time for the assessment does not exceed tred,
original crack plus the amount of creep crack growth. then this simpli®ed treatment of transient creep is not
The crack tip parameter is estimated using reference adequate and it is necessary to use the parameter C(t)
stress techniques as explicitly in estimating creep crack growth.

C p ˆ s ref e_ cref R 0 7† 2.1. Treatment of weldments

where e_ cref is the creep strain rate from uniaxial data at the The basic R5 procedures described above for homo-
reference stress (Eq. (1)) calculated for the current defect geneous components can be re®ned to take account of stress
size a. The parameter C p characterises the crack tip stress redistribution between the different material zones of a
and strain rate ®elds for times in excess of the redistribution weldment due to the differing creep deformation behaviour
time. of the parent material, weld, and HAZ. This typically occurs
Prior to the attainment of widespread creep conditions, in pressurised butt-welded pipes, where the maximum
the crack tip stress and strain rate ®elds are characterised by principal (hoop) stress is parallel to the weld, in the absence
a parameter usually denoted C t†: For times in excess of the of additional axial loadings. This leads to the R5 k-factor,
redistribution time C t† approaches C p. An interpolation where k , 1 or k . 1 for zones relatively weak or strong in
formula for C t† during the transition between initial elastic creep compared with the parent material. For loading predo-
loading and steady state secondary creep has been derived minantly transverse to the weld, such overall stress
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 965

Fig. 1. 12 CrMoV normalised and tempered: comparison of estimated C p Cref

p p
† with the value (Cexp ) deduced from experimental measurements.

redistribution cannot occur and k ˆ 1: The homogeneous 3. Validation of the basic R5 defect assessment
reference stress of Eq. (1) is multiplied by k for each procedure
characteristic material zone within the weld and Eq. (2) is
then used to assess rupture of each zone using the The basic R5 defect assessment procedure has been
appropriate local rupture data. Crack growth follows from validated by considering:
Eqs. (7) and (11) or (10) and (12), again using the relevant
factored reference stress and creep crack growth and (a) particular aspects of the procedure individually, or
deformation data appropriate to the cracking location. Strain (b) the procedure as a whole.
compatibility under hoop dominated loading in pipes,
however, enables the creep strain rate in Eq. (4) to be set This section gives examples of both types of validation.
to the parent rate in that case.
3.1. Validation of reference stress estimates of C p and C t†
2.2. Creep±fatigue crack growth
A feature of the R5 defect assessment procedure is the use
Under conditions of shakedown to linear elastic of reference techniques to estimate C p using Eq. (7).
conditions, the total crack growth per cycle is given by Validation of this aspect of the procedure has been
the sum of creep crack growth calculated using the addressed by comparing reference stress estimates of C p
approaches described above and fatigue crack growth. An with experimentally derived values from laboratory creep
allowance is made for compressive stresses at the extreme crack growth tests and with ®nite-element results.
of the cycle by replacing the stress intensity factor range DK Reference stress estimates have been shown to give good
by DKeff ˆ q0 DK; where q0 # 1 is a function of or conservative predictions of experimentally derived C p
R ˆ Kmin =Kmax , in a modi®ed Paris Law to de®ne the fatigue values for a wide range of steels; Figs. 1 and 2 show typical
crack growth per cycle, (da/dN)f: results for 12 CrMoV parent and simulated HAZ materials
q0 ˆ ‰1 2 0:5 RŠ=‰1 2 RŠ 14† respectively.
In addition, the reference stress technique for estimating
da=dN†f ˆ C DKeff †m 15† C p has been compared with ®nite-element results for a range
of geometries in the EPRI/GE scheme [11]. The EPRI/GE
where C and m are constants. Interaction between the two scheme contains tables, which represent normalised ®nite
modes of crack growth is admitted by incorporating the element J solutions for power-law plastic materials. These
effects of creep damage in the constant C of Eq. (15) via a may also be used to calculate C p for secondary creep beha-
dependence on dwell time. viour when creep rates are de®ned by Eq. (9). Miller and
966 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

Fig. 2. 12 CrMoV untempered coarse grained Bainite: comparison of estimated C p Cref

p p
† with the value (Cexp ) deduced from experimental measurements.

Ainsworth [12] have compared reference stress estimates of and a creep stress index of 5 was used throughout the
C p given by Eq. (7) with the EPRI/GE ®nite-element studies, with changes in temperature modelled by changes
solutions for a range of ¯awed geometries and creep stress in the constant D in Norton's law (Eq. (9)). Results of these
indices. The reference stress estimates were shown to analyses were accurately predicted using the estimation
accurately reproduce the ®nite-element results for both formula of Eq. (10) as shown in Fig. 6.
test specimen and cylindrical geometries. On average the In summary, reference stress techniques have been shown
estimates correspond to a conservatism of about 5% on load to provide good or conservative predictions of the crack tip
but may vary by about ^15%. Non-conservatism relative to parameters C p and C t† based on the results of both experi-
the ®nite-element solutions occurs mainly for small cracks mental and ®nite-element validation.
in tension geometries at low stress index values. However,
any slight non-conservatism in the reference stress estimates 3.2. Validation of the assessment procedure
of C p for such geometries will not introduce non-conserva-
tism in practical assessments using the procedures of Validation of the overall R5 creep crack growth assess-
Section 2 as a result of other inherent pessimisms (e.g. the ment procedures is illustrated here for a defect in a large
use of creep crack growth data obtained from deeply 1 1 1
2 Cr 2 Mo 4 V pressure vessel.
cracked bend geometries). The cylindrical vessel of outer radius R0 ˆ 175 mm and
The ®nite-element validation has also been extended to thickness tˆ 60 mm was constructed using 12 Cr 12 Mo 14 V
cover the redistribution period where it has been shown that steel pipe and 12 Cr 12 Mo 14 V forged end-caps connected by
good or conservative estimates of the parameter C t† are a number of 2Cr1Mo multipass butt welds (Fig. 7). A
obtained using Eqs. (8) or (10) for a range of geometries special welding procedure produced weld fusion boundaries
and both primary and combined primary and secondary and hence heat affected zone (HAZ) and Type IV regions,
loadings (Figs. 3±5) based on idealised elastic-power law which were normal to the vessel surface. A number of exter-
creep behaviour. Finite-element studies of the effect of nal part circumferential defects were spark machined into
temperature changes, occurring both during the transition the weld HAZs. The pressure vessel was then tested at a
phase and under steady state creep conditions, have been temperature of 5658C and internally pressurised with steam
reported [13] for a centre cracked plate (CCP) geometry at 25 MPa for the ®rst 2000 h and 35 MPa for the remainder
with a=w ˆ 0:25: Plane strain conditions were assumed of the test period. A brief summary of an assessment of
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 967

C(t)/C* 5

4 Finite Element

Plane Strain Reference Stress

3 Approximation

0
0 1 2 3 4 5 6 7 8
Ratio of Reference Creep to Elastic Strain

Fig. 3. Comparison of estimates of C t†=C p with ®nite element results for CCP specimen with a=w ˆ 0:2 subjected to load controlled tension.

creep crack growth from a defect, 16 mm deep by 160 mm circumferential defect in a pressurised pipe is that for an
long, located on the pipe side of the stress-relieved end-cap uncracked cylinder under pressure:
closure weld (Fig. 7) is given here. Further details of this
s ref ˆ p=ln R0 =Ri † 16†
assessment and assessments of other defects in this test
vessel have been published previously [14,15]. The initial value of s ref ˆ 83:4 MPa and this is unchanged
As the purpose of this section is to provide validation of throughout the assessment. p The
 initial value of stress inten-
the R5 procedures for predicting creep crack incubation and sity factor is K ˆ 6:6 MPa m; which results in an initial
growth, explicit details of associated creep rupture and time value of the length parameter R 0 ; from Eq. (6) of
independent R6 [6] fracture calculations are omitted for
R 0 ˆ 6:2 mm 17†
brevity. To assess creep crack incubation and growth
according to the basic R5 procedure of Section 2, the defect Using an incubation COD for the coarse grained HAZ mate-
has been idealised as a 10:1 aspect ratio semi-elliptical rial of di ˆ 0:01 mm results in a value for the corresponding
external surface defect. Details of the stress intensity factor reference strain from Eq. (5)
and reference stress solutions assumed in the current calcu-
ec ˆ 0:005 18†
lations are given in Ref. [14]. However, it is worth noting
that the appropriate reference stress solution for a shallow The incubation time is then obtained as the time to

10

7
Finite Element
6
C(t)/C*

Reference Stress
Approximation
5

0
0 1 2 3 4 5 6
Ratio of Reference Creep to Elastic Strain

Fig. 4. Comparison of estimates of C t†=C p with ®nite element results for ECCC specimen with a=w ˆ 0:25 subjected to internal pressure.
968 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

10

7 Finite Element

6 Reference Stress
C(t)/C* Approximation

0
0 1 2 3 4 5 6 7 8 9
Ratio of Reference Creep to Elastic Strain

Fig. 5. Comparison of estimates of C t†=C p with ®nite element results for ECCC specimen with a=w ˆ 0:25 subjected to combined internal pressure and
thermal stress (equal pressure and thermal stress intensity factors).

accumulate the above strain using appropriate creep microstructure, it is possible to assess the crack growth
deformation data pertaining to the reference stress and behaviour using either
temperature. This gives a predicted incubation time of
28 000 h. (a) different crack growth laws for the coarse and re®ned
The creep strain of Eq. (18) is in excess of the elastic HAZ regions
strain at the reference stress, so that widespread creep condi- a_ c ˆ Ac Cp †q 19†
tions are established prior to incubation (i.e. ti . tred ).
Therefore, the redistribution period does not have to be
a_ R ˆ AR C p †q 20†
considered in estimating the subsequent creep crack growth,
which is estimated using Eq. (11) together with reference or
stress estimates of C p from Eq. (7). However, in this parti- (b) a weighted average creep crack growth law which
cular example where the defect is located in a HAZ takes account of the ratio of coarse to re®ned microstruc-
comprising alternate regions of coarse and ®ne-grained ture, denoted a . The weighted average creep crack

10

8
Isothermal

7
Increase in Temp. at Steady State

6 Decrease in Temp. at Steady State

C(t)/C*

5 Increase in Temp. During Redistribution Period

Decrease in Temp. During Redistribution Period

4
Plane Strain Reference Stress Approximation
3 (Equation A7.19)

0
0 1 2 3 4 5 6 7 8 9
Ratio of Reference Creep to Elastic Strain

Fig. 6. Comparison of estimates of C t†=C p with ®nite element results for CCP specimen with a=w ˆ 0:25 subjected to temperature changes under pure
mechanical loading.
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 969

Fig. 7. Schematic of pressure vessel showing welds and location of HAZ defect.

growth rate a_m ; is given by: 4. Use of the R5 procedures in plant assessment

a_m ˆ Am C p †q 21† The R5 defect assessment procedures are widely used to

assess components in British Energy's Advanced Gas
Cooled Reactor (AGR) plant. In this section the use of R5
Ac 1 1 a † is illustrated by calculating the creep±fatigue crack growth
Am ˆ 22†
Ac =AR † 1 a in a speci®c component.

4.1. Component details

Predicted creep crack growth using instantaneous and The component selected for this illustration is an AGR
weighted average crack growth rates are shown in reheater outlet penetration, for which a simpli®cation of the
Figs. 8(a) and (b), respectively, together with ultrasonic geometry is shown in Fig. 9. Essentially, the concrete vessel
measurements of crack depth obtained during test interrup- is insulated from the steam tube by means of a sheath tube,
tions. In both cases, close agreement is obtained between the which is connected to the steam tube by means of a forging.
R5 predictions and experimental observations. In practice, the steam tube is thicker than its normal design
970 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

Fig. 8. R5 calculations of crack growth for pressure vessel defect using both instantaneous and mean creep crack growth laws.

thickness and this leads to a `notch' where the tube meets The forging and steam tube are made of Type 316H
the forging which has the same thickness as the nominal stainless steel and are joined by Type 316 weld metal.
steam tube. Relevant dimensions are given on Fig. 9. Based on plant operating conditions, a mean effective
One of these components has recently been replaced temperature for assessing creep crack growth is 5208C
following the discovery of a defect at the location indicated whereas the maximum temperature to be used in fracture
in Fig. 9. From the subsequent failure investigation it is assessments is 5808C. Data for these steels at these tempera-
understood that a reheat crack developed to a depth of tures were available from handbooks within British Energy,
8 mm and a surface length of 137 mm, at a time of approxi- and the properties were con®rmed by some testing of
mately 50 000 h. This ¯aw had grown to become fully pene- ex-service material.
trating by the time the defect was detected as a steam leak at The steam tube operates under a steady internal pressure
,130 000 h. For the purposes of these calculations, the defect of 3.9 MPa. This pressure is used for calculations of creep
is idealised as semi-elliptical with an aspect ratio of 8.6:1. and fatigue crack growth below. To determine a limiting
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 971

Fig. 9. Geometry and dimensions of AGR penetration component.

defect size, however, a nominal fault pressure of 4.6 MPa is The results of the creep±fatigue crack growth calcula-
used. tions are shown in Fig. 10 based on
In addition to pressure loading, the component is
subjected to global bending from temperature differences (i) mean creep deformation and mean creep crack growth
across the sheath tube and from system loadings from data, and
adjoining pipework. The total global bending moment is (ii) lower bound creep deformation and upper bound
75.5 kNm. Welding residual stresses as modi®ed by the creep crack growth data.
proof pressure test and subsequent operation at elevated
temperature are also considered. These are two of the three combinations of creep defor-
mation and creep crack growth properties recommended for
4.2. Assessment details use in R5 [3] in recognition of the inverse correlation
between creep deformation rates and creep crack growth
A limiting defect depth of 13.5 mm has been calculated rates. Experience has shown that for Type 316H material,
using R6 [6] with lower bound material properties assuming the third recommended combination of upper bound creep
a 8.6:1 aspect ratio. For conservatism, this is based on the deformation and lower bound creep crack growth rates
stresses at the most highly stressed location around the generally results in the lowest predictions of crack growth.
circumference (the top of the penetration). The crack growth calculations include a fatigue contribu-
Prior to the crack growth calculations, an assessment is tion resulting from plant start-up/shutdown cycles. This
made of continuum damage by comparing the reference contribution has been evaluated using either
stress with creep rupture data (see Eq. (2)). For the
uncracked penetration, and for the 8 mm deep, 8.6:1 aspect (a) pure fatigue crack growth data, or
ratio defect, the continuum damage at 130 000 h was found (b) enhanced fatigue crack growth data derived using the
to be insigni®cant so that creep rupture is not limiting. As cyclic component of crack growth from creep-fatigue
required by R5, these calculations were repeated taking tests which include dwell periods.
account of in-service growth but this did not affect the
conclusion that creep rupture in the ligament ahead of the As the component is operating at a relatively low
defect is not limiting. temperature (5208C), stress redistribution is not complete
Creep±fatigue crack growth calculations have been within the assessment period and therefore C t† rather
performed for a 8 mm deep 8.6:1 ratio circumferential than C p is used throughout the creep crack growth calcula-
defect using the basic R5 procedures, as summarised in tions. This in turn means that the reference creep strains
Section 2. In the continuum damage calculations above, accumulated during the assessment period will be small
creep damage was conservatively calculated assuming the and it is considered that use of the enhanced fatigue crack
crack was present from the start-of-life. For the creep± growth data derived from creep±fatigue tests under
fatigue crack growth calculations it was assumed that the widespread creep conditions will result in overly conserva-
crack had just formed at the current operating time, so that tive predictions of crack growth rates. This is con®rmed by
crack tip stresses, as characterised by C t† were maximised. comparing the predictions given in Fig. 10 and is also
In addition, it was assumed that prior creep strain was consistent with the metallurgical examination of the compo-
negligible so that creep strain rates at the reference nent which concluded that the cracking was dominated by
stress were calculated as starting from the current creep processes [16]. The results based on a pure fatigue
operating time. crack growth law presented in Fig. 10 are therefore
972 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

Fig. 10. Results of R5 creep±fatigue calculations for defective AGR penetration component.

considered to be most representative for the present assess- growth law and creep crack growth estimated using C t†
ment. For each combination of creep deformation and crack (allowing for a short incubation period) together with
growth properties, calculations are shown in Fig. 10 with recommended combinations of creep deformation and
and without an incubation period evaluated using the s d creep crack growth properties.
approach [17]. It is evident that inclusion of an incubation
period results in a lower initial value of C(t) being used in
the crack growth calculations and therefore reduced initial 5. Development of alternative methods
crack growth rates are obtained. This pragmatic approach of
including an incubation time for a defect which had formed The methods for assessing incubation and the early stages
as a result of creep mechanisms has been adopted to re¯ect of creep crack growth described in Section 2 are based on
the fact that the creep strain accumulated during formation the evaluation of parameters including crack opening
of the reheat crack would result in a ®nite initial value of displacement d , and crack tip parameters C p and C t†
C t†: together with experimental data describing creep crack
The base case creep±fatigue crack growth calculations initiation or growth. However, for low temperature fracture,
shown in Fig. 10, which include an incubation period, the simpli®ed R6 procedure [6] has been developed, which
show that a good but conservative prediction of the time uses the concept of a Failure Assessment Diagram (FAD) to
for the defect to reach a critical depth is obtained for the avoid detailed calculations of crack tip parameters. In recent
combination of lower bound creep deformation and upper years, FAD approaches have been extended to the creep
bound creep crack growth data. However, if mean data are regime [18±21] and the high temperature Time Dependent
used, a non-conservative prediction of crack growth Failure Assessment Diagram (TDFAD) method has been
behaviour is obtained. For the situation being analysed formally incorporated into R5 [3]. This section ®rst
here, that is a failure of a single component in a population describes the stress-based TDFAD approach currently in
of 48, it is reasonable to expect that it may be necessary to R5 and then brie¯y describes progress on development of
use bounding materials properties data to predict the an alternative strain-based TDFAD approach.
observed behaviour; the behaviour of the remaining The stress-based TDFAD is based on the Option 2 FAD in
components, which have not failed, would be expected to R6 [6] and involves a failure assessment curve relating the
be more representative of mean properties. two parameters Kr and Lr, de®ned in Eqs. (23) and (24)
The calculations presented in this section have illustrated below, and a cut-off Lmax
r : For the simplest case of a single
the use of the R5 procedures for predicting creep±fatigue primary load acting alone
crack growth in an AGR component. A conservative predic- c
Kr ˆ K=Kmat 23†
tion of the observed crack growth behaviour was obtained
c
using the R5 procedures together with a pure fatigue crack where K is the stress intensity factor and Kmat is the
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 973

Stress

0.2 Stress-Strain Curve (t=0)

e
ref
ref
e Isochronous Curve (t>0)
0.2
c
0.2

0.002 Total Strain

εref

Fig. 11. Schematic isochronous stress±strain curve.

appropriate creep toughness value, and In Eq. (25), E is Young's modulus and eref is the total strain
c from the average isochronous stress±strain curve at the
Lr ˆ s ref =s 0:2 24†
c
c
where s ref is the reference stress and s 0:2 is the stress reference stress s ref ˆ Lr s 0:2 ; for the appropriate time and
corresponding to 0.2% inelastic (plastic plus creep) strain temperature. Thus, Eq. (25) enables the TDFAD to be
from the average isochronous stress±strain curve for the plotted with Kr as a function of Lr, as shown schematically
temperature and assessment time of interest, see Fig. 11. in Fig. 12. The cut-off Lmax
r is de®ned as:
The FAD is then de®ned by the Eqs. (25) and (26) Lmax c
ˆ s R =s 0:2 27†
r
" #21=2
Eeref L3r s 0:2
c
where s R is the rupture stress for the time and temperature
Kr ˆ c 1 Lr # Lmax
r 25†
Lr s 0:2 2Eeref of interest. For consistency with R6 [6] at short times, the
value of Lmax
r should not exceed s =s 0:2 where s is the short
Kr ˆ 0 Lr . Lmax
r 26† term ¯ow stress and s 0:2 is the conventional 0.2% proof

1.0

R6 Option 1

0.8 t=0

t = 3000 h

0.6 t = 300,000 h
Kr

0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Lr

Fig. 12. Schematic FAD based on data from an austenitic steel.

974 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

Fig. 13. Schematic load±displacement behaviour from a constant load creep crack growth test.

stress. As in R6 [6] s may be taken as s 0:2 1 s u †=2 where displacement curve

s u is the ultimate tensile strength. hU T
A central feature of the TDFAD approach is the de®nition JT ˆ 31†
c B n w 2 a0 †
of an appropriate creep toughness Kmat which, when used in
conjunction with the FAD, ensures that crack growth in the where w is specimen width, a0 is initial crack length, Bn is
assessment period is less than a value Da: Creep toughness net specimen thickness and
values may be estimated indirectly from conventional creep h ˆ 2 1 0:522 1 2 a0 =w† 32†
crack initiation and growth data in the form of Eqs. (4) and
c
(11), respectively, as for CT specimens. Values of creep toughness Kmat ; may then
be derived from creep crack growth tests as a function of
" #1=2
E B 1 Da=A†1=q crack growth increment Da, using
c
Kmat ˆ 28† p
t 1=q21† c
Kmat ˆ E 0 JT 33†

Alternatively, direct approaches for determining creep in conjunction with Eq. (31). Fig. 14 shows schematically
toughness based on experimental load±displacement data the variation of creep toughness with time at initiation,
can be based on methods used to derive the J-integral and which is de®ned for engineering purposes as corresponding
hence the fracture toughness, given in low temperature to 0.2 mm crack extension.
fracture toughness standards such as the ESIS procedure The point Lr ; Kr †; from Eqs. (23) and (24) using the
[22]. Consider a load-controlled creep crack growth test current values of stress intensity factor and reference stress,
conducted on a standard compact tension (CT) specimen respectively, is plotted on the FAD illustrated in Fig. 12. If
resulting in a typical load±displacement trace of the form the point lies within the failure assessment curve of Eq. (25)
shown in Fig. 13. If it is assumed that the amount of crack and the cut-off de®ned by Eqs. (26) and (27), then the crack
growth in the test Da is small, the total displacement DT may extension is less than Da and creep rupture is avoided. Alter-
be conveniently partitioned into elastic, plastic and creep natively, the TDFAD approach can be used to predict the
components, denoted De ; Dp and Dc ; respectively, where time required for the crack to extend by Da: This requires
the time locus of Lr ; Kr † points to be constructed and the
DT ˆ De 1 Dp 1 Dc 29† time for crack extension Da; is given by the intersection of
this locus with the failure assessment curve of Eq. (25) for
Similarly, the total area under the load±displacement curve the corresponding time. This calculation may be simpli®ed
UT, may be conveniently partitioned into elastic, plastic and by noting that the failure assessment curve is a weak func-
creep components, denoted Ue, Up and Uc respectively: tion of time [20]; this allows the time for the crack to extend
by Da to be estimated using a failure assessment curve for a
UT ˆ Ue 1 Up 1 Uc 30† single time.
A strain-based TDFAD can also be derived based on the
The ESIS fracture toughness testing procedure [22] Option 2 FAD speci®ed in R6 [6] or, alternatively, by
evaluates experimental total J values JT, using the following combining Eqs. (24) and (25) above. This strain-based
relationship based on the total area under the load± approach is currently being developed [23], and is similar
 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976 975

Fig. 14. Schematic creep toughness data.

to the stress-based approach described above except that the [2] Webster GA, Ainsworth RA. High temperature component life
ratio of total to elastic reference strains is used instead of Lr assessment. London: Chapman & Hall, 1994.
[3] British Energy Generation Ltd. An assessment procedure for the high
as the abscissa of the FAD.
temperature response of structures, R5 Issue 2 Revision 2, 1998.
[4] PD6539: 1994. Guide to methods for the assessment of the in¯uence
of crack growth on the signi®cance of defects in components operat-
6. Concluding remarks ing at high temperatures. London: BSi, 1994.
[5] E1457-98. Standard test method for measurement of creep crack
This paper has described the basic R5 defect assessment growth rates in metals. Philadelphia: American Society for Testing
procedure for predicting creep crack growth in homoge- and Materials, 1998.
[6] British Energy Generation Ltd. Assessment of the integrity of struc-
neous components. In addition: tures containing defects. R6 Revision 3, Amendment 11, 2000.
[7] Riedel H, Rice JR. Tensile cracks in creeping solids: Fracture
1. Extensions to the procedure to incorporate the effects of mechanics: Twelfth conference. ASTM STP 700, 1980. p. 112±30.
fatigue and the treatment of weldments have been [8] Ainsworth RA. Initiation of creep crack growth. Int J Solid Struct
discussed. 1982;18:873±81.
[9] Schwalbe KH, Ainsworth RA, Saxena A, Yokobori T. Recommenda-
2. Examples of validation of the procedure as a whole and tions for a modi®cation of ASTM E1457 to include creep-brittle
of the particular aspect of use of reference stress tech- materials for Engng Fract Mech 1999;62:123±42.
niques to estimate C p and C t†; have been described. [10] Ainsworth RA, Budden PJ. Crack tip ®elds under non-steady creep
3. Use of the R5 procedure has been illustrated by applica- conditions Ð I. Estimates of the amplitude of the ®elds. Fatigue Fract
tion to the prediction of creep±fatigue crack growth from Engng Mater Struct 1990;13:263±76.
[11] Kumar V, German MD, Shih CF. An engineering approach for elastic
a defect in an Advanced Gas Cooled Reactor component.
plastic fracture. EPRI Report NP-1931, 1981.
4. Development of alternative TDFAD methods for predict- [12] Miller AG, Ainsworth RA. Consistency of numerical results for
ing incubation and the early stages of creep crack growth power-law hardening materials and the accuracy of the reference
have been described, including some aspects relating to stress approximation. J Engng Fract Mech 1989;32:233±47.
the determination of the required creep toughness [13] Budden PJ, Dean DW, Turner GE. Creep crack growth prior to stress
parameter. redistribution. Mechanics of creep brittle materials, vol. 2. London:
Elsevier, 1990. p. 90±9.
[14] Jones MR, Coleman MC. The assessment of creep crack growth in a
welded pressure vessel. Proceedings of the Fourth International
Acknowledgements Conference on Creep and Fracture of Engineering Materials and
Structures, Swansea: Institute of Metals, 1990. p. 605±19.
This paper is published with permission of British Energy [15] Budden PJ. Analysis of type IV creep failures of three welded ferritic
Generation Ltd. pressure vessels. Int J Pres Ves Pip 1998;75:509±19.
[16] Dunn JW, McGuigan J, McLean RI, Miles L, Stevens RA. Inves-
tigation and repair of a leak at a high temperature stainless steel
butt weld. Proc Int Conf Integrity High Temp Welds, Nottingham
References 1998:241±58.
[17] Moulin D, Drubay B, Laiarinandrasana LA. Practical method based
[1] Ainsworth RA, Chell GG, Coleman MC, Goodall IW, Gooch DJ, on stress evaluation (s d criterion) to predict initiation of crack under
Haigh JR, Kimmins ST, Neate GJ. CEGB assessment procedure for creep and creep fatigue conditions. ASME J Pres Ves Tech
defects in plant operating in the creep range. Fatigue Fract Engng 1995;117:335±40.
Mater Struct 1987;10:115±27. [18] Ainsworth RA. The use of a failure assessment diagram for initiation
976 D.W. Dean et al. / International Journal of Pressure Vessels and Piping 78 (2001) 963±976

and propagation of defects at high temperatures. Fatigue Fract Engng [21] Ainsworth RA, Hooton DG, Green D. Failure assessment diagrams
Mater Struct 1993;16:1091±108. for high temperature defect assessment. Engng Fract Mech
[19] Hooton DG, Green D, Ainsworth RA. An R6 type approach for the 1999;62:95±109.
assessment of creep crack growth initiation in 316L stainless steel test [22] European Structural Integrity Society. ESIS procedure for determin-
specimens. Proc ASME PVP Conf, Minneapolis 1994;287:129±36. ing the fracture behaviour of materials, ESIS P2-92, 1992.
[20] Ainsworth RA, Hooton DG, Green D. Further developments of an R6 [23] Smith DJ, Fookes AJ, Dean DW, Lamb M. Development of a strain-
type approach for the assessment of creep crack incubation. Proc based FAD for prediction of creep cracking in welds. Proc Int Conf
ASME PVP Conf, Honolulu 1995;315:39±44. Integrity High Temp Welds, Nottingham 1998:355±70.